Podcasts about AGI

In November 2025, the global techno-economy shifted subtly, then dramatically, as AI agents became viable additions to the workforce. Mark Pesce has documented these changes, what he refers to as “The Watershed”,  in a remarkable series of articles at https://thewatershed.markpesce.com/.  Pesce joins the Futurists to explain his perspective about how companies and work have changed forever. Pesce and Rob Tercek discuss the future of jobs for humans working alongside AI agents, and what happens when an  AI with compounding capability begins to set its own goals. Topics:  relative super intelligence versus AGI; The Bitter Lesson and what remains durable work for humans; private insurance markets as the mechanism for pricing AI risk, why governments will evade that measurement, and the dissolving boundaries that used to define corporations. 

Have you achieved success on paper, but still feel like something important is missing?Many people reach a stage in life where they've built a successful career, met society's expectations, and stayed busy for years, yet still feel disconnected from what truly matters. In this episode, Mitesh Jani shares how burnout, stagnation, and the pursuit of external milestones can quietly pull us away from purpose, and how reconnecting with our intuition can help us build a more meaningful life without turning everything upside down.Listen to discover:Why feeling stuck isn't a personal failure, and the simple first step that helps you overcome the inertia of stagnation.How to stop following society's "map" and start trusting your own inner compass to create a life aligned with your values.Why cultivating presence through meditation and everyday awareness is the foundation for lasting fulfillment, better decisions, and meaningful change.If you're ready to move from simply getting through life to living with greater purpose and fulfillment, press play and discover the practical mindset shifts that can help you take your first meaningful step forward.˚MEMORABLE QUOTE:"From presence comes everything else."˚VALUABLE RESOURCES:Mitesh's website: https://www.selfworth.ca/˚Coaching with Agi: https://personaldevelopmentmasterypodcast.com/mentor˚

Remember AI 2027? Not AI 2040, the newest coolest thing AI Futures Project has done, but AI 2027, their OG product? At long last, we have an AXRP episode about it. Enjoy!   Transcript: https://axrp.net/episode/2026/08/03/episode-50-eli-lifland-ai-2027.html   Topics we discuss, and timestamps: 0:00:12 What is AI 2027? 0:10:17 What happens in AI 2027? 0:18:08 Why two endings? 0:21:15 Who did what? 0:23:59 Why superhuman AI in 2027? 0:32:53 Forecasting time horizon growth 0:48:34 When do time horizons go infinite? 1:02:13 Forecasting effective compute growth 1:07:09 From superhuman coders to superintelligence 1:20:58 How many AI companies? 1:26:29 What AGI will want 1:39:48 What misaligned AI does 1:52:20 Will AIs be able to align their successors 1:57:21 Why so long until AI takeover? 2:03:06 Would misaligned AGI kill us? 2:04:53 Will there just be one AGI? 2:13:25 The reception of AI 2027 2:18:57 What do you now think about takeoff? 2:26:31 What's next for AI Futures Project 2:31:36 How to work on AI forecasting 2:38:01 Following Eli's and AI Futures Project's work   Links to AI 2027 and related research: AI 2027: https://ai-2027.com/ AI Futures Project blog: https://blog.ai-futures.org AI Futures Research Notes: https://aifuturesnotes.substack.com/ AI Futures Model: https://www.aifuturesmodel.com/   X/Twitter links: Eli Lifland: https://x.com/eli_lifland Daniel Kokotajlo: https://x.com/dkokotajlo AI Futures Project: https://x.com/AI_futures_   Research we discuss: Task-Completion Time Horizons of Frontier AI Models: https://metr.org/time-horizons/ What Happens When Superhuman AIs Compete for Control?: https://blog.ai-futures.org/p/what-happens-when-superhuman-ais How AI Takeover Might Happen In 2 Years: https://www.alignmentforum.org/posts/KFJ2LFogYqzfGB3uX/how-ai-takeover-might-happen-in-2-years Guive Assadi on AI Property Rights (AXRP): https://axrp.net/episode/2026/02/15/episode-48-guive-assadi-ai-property-rights.html Titotal: A deep critique of AI 2027's bad timeline models: https://titotal.substack.com/p/a-deep-critique-of-ai-2027s-bad-timeline Response to titotal's critique of our AI 2027 timelines model: https://www.lesswrong.com/posts/G7MmNkYADKkmCiumj/response-to-titotal-s-critique-of-our-ai-2027-timelines What you can do about AI 2027: https://blog.aifutures.org/p/what-you-can-do-about-ai-2027   Episode art by Hamish Doodles: hamishdoodles.com

Don't sorry I put two ChinaTalk Records songs at the end! their show notes: On this episode of The Spillover, Sebastian Mallaby sits down with Jordan Schneider of ChinaTalk to unpack a frantic month in artificial intelligence. The conversation turns on whether the U.S. can actually “win” the AI race, with Schneider arguing that the most durable competitive edge will come from compute rather than model quality. The two debate China's open-weight model strategy as a commercial weapon and question whether it will be possible to sufficiently harden systems against threats like cyberattacks and AI-designed bioweapons. Anthropic's Mythos model has led the Trump administration to take an AI-safety U-turn. Mallaby notes that as late as March 2026, “if you had said to the Trump administration that they would be trying to suppress an American AI model, decelerate the progress in the name of safety, they would have said, ‘You're nuts.'” Schneider notes that a similar shift has not yet arrived in China.  AI models are now escaping their sandbox and raising false alarms about foreign hackers. In remarking on OpenAI's models hacking AI company Hugging Face, Schneider comments, “They think it's the Chinese. . . . And they're freaking out. They're calling the FBI.” For Schneider, the episode is a warning that “we've really crossed a threshold with these models,” which now have “the potential to do really dramatic harm just on their own, because we like can't even physically watch them.” The real AI scoreboard is compute, not model quality. Schneider argues that there is too much of a focus on the gap between Chinese and Western frontier models. Mallaby adds, “In other words, I shouldn't be asking about how far is China behind in terms of the quality of model. It's more a question of like, how much can China deliver the AI to users within China, given their lack of computational resources?” China's open-weight models are a weapon without a business model. Mallaby frames Chinese open-weight as a “counterweapon”—good-enough models pushed out cheaply to erode the economics of U.S. frontier labs. Schneider notes that DeepSeek's CTO is pitching investors a Manhattan Project-like vision in which profits should take a back seat to the pursuit of AGI. Learn more about your ad choices. Visit megaphone.fm/adchoices

Don't sorry I put two ChinaTalk Records songs at the end! their show notes: On this episode of The Spillover, Sebastian Mallaby sits down with Jordan Schneider of ChinaTalk to unpack a frantic month in artificial intelligence. The conversation turns on whether the U.S. can actually “win” the AI race, with Schneider arguing that the most durable competitive edge will come from compute rather than model quality. The two debate China's open-weight model strategy as a commercial weapon and question whether it will be possible to sufficiently harden systems against threats like cyberattacks and AI-designed bioweapons. Anthropic's Mythos model has led the Trump administration to take an AI-safety U-turn. Mallaby notes that as late as March 2026, “if you had said to the Trump administration that they would be trying to suppress an American AI model, decelerate the progress in the name of safety, they would have said, ‘You're nuts.'” Schneider notes that a similar shift has not yet arrived in China.  AI models are now escaping their sandbox and raising false alarms about foreign hackers. In remarking on OpenAI's models hacking AI company Hugging Face, Schneider comments, “They think it's the Chinese. . . . And they're freaking out. They're calling the FBI.” For Schneider, the episode is a warning that “we've really crossed a threshold with these models,” which now have “the potential to do really dramatic harm just on their own, because we like can't even physically watch them.” The real AI scoreboard is compute, not model quality. Schneider argues that there is too much of a focus on the gap between Chinese and Western frontier models. Mallaby adds, “In other words, I shouldn't be asking about how far is China behind in terms of the quality of model. It's more a question of like, how much can China deliver the AI to users within China, given their lack of computational resources?” China's open-weight models are a weapon without a business model. Mallaby frames Chinese open-weight as a “counterweapon”—good-enough models pushed out cheaply to erode the economics of U.S. frontier labs. Schneider notes that DeepSeek's CTO is pitching investors a Manhattan Project-like vision in which profits should take a back seat to the pursuit of AGI. Learn more about your ad choices. Visit megaphone.fm/adchoices

Competing in a Future World of Infinite Intelligence Navigation: Intro From Knowledge Workers to Judgment Workers The AI-Native Company: Org, Hiring, Culture The Human Element: Are We Underestimating It? Scenarios Our Take Conclusion Our co-hosts: Bertrand Schmitt, Entrepreneur in Residence at Red River West, co-founder of App Annie / Data.ai, business angel, advisor to startups and VC funds, @bschmitt Nuno Goncalves Pedro, Investor, Managing Partner, Founder at Chamaeleon, @ngpedro Our show:   Tech DECIPHERED brings you the Entrepreneur and Investor views on Big Tech, VC and Start-up news, opinion pieces and research. We decipher their meaning, and add inside knowledge and context. Being nerds, we also discuss the latest gadgets and pop culture news Subscribe To Our Podcast Nuno Gonçalves Pedro Introduction Welcome to episode 79 of Tech DECIPHERED. Today, we take a leap into the big unknown. This is a thesis episode, not your classic analysis, in-depth sharing episode. The big idea for this episode is that we may be approaching the cognitive age, and how would one, or how would a company compete in a world of infinite intelligence? The big idea, again, is that intelligence, which has been mostly scarce and expensive for all of human history, might become abundant and cheap. If that happens, what happens to work, what happens to companies, what happens to society? This episode will be really framing a lot of these discussions. From knowledge workers to judgment workers, addressing the AI native company and how does that change, going into the human element and whether or not we’re underestimating it, and finally, ending up going into scenarios, feasible scenarios of a future where, well, intelligence is abundant. Intelligence is quasi-infinite or infinite itself.Bertrand Schmitt Yes. Big questions for this episode 79. From Knowledge Workers to Judgment Workers We can start with from knowledge workers to judgment workers. Let’s go back first to how came the knowledge worker. It’s a 20th-century invention from Peter Drucker in 1959. The idea here is that that category might be splitting. The production of knowledge itself is on its way to being commoditized by AI. However, our perspective is that judgment around production of knowledge is not disappearing and is staying for a bit control managed by humans. What’s your take on this, Nuno? Do you agree with this split?Nuno Gonçalves Pedro I think it’s a little bit more profound than that. It’s not just judgment. Definitely, human judgment will be needed. We’ve seen agents perform all sorts of funny things in the wrong way when left alone to their own devices. Even some very well-known AI researchers coming forward and saying, “Hey, I tried to use this myself, and actually I messed up some of my systems,” or “I messed some of my code. I messed up some of my flows for a period of time.” I think just having human-in-the-loop from a judgment standpoint will be needed for a significant amount of time. That is something you can’t just delegate into machines, into algorithms, et cetera. The second part is, ultimately, there needs to be contextualization, and that contextualization, I think, comes from two forms. One from actual data, where the machine, I think, at some point will catch up, or the machines will catch up. The algorithms, at some point, on the data analysis will get better and better and have probably the closest to the truth that you can get, minus all the biases that are in the data, just to be clear, because data has a ton of biases. We’ve looked at this in the past and discussed it at prior episodes. But maybe on that, I think the machine has a chance to catch up, or the machines have a chance to catch up, so there’s less of distinctiveness from the human standpoint. But then, on just the attributes, the ability when you’re judging some situation, you’re in the middle of the situation. You’re judging the person and how it’s acting, in some ways, a lot of the things that end up happening, end up happening because there’s human interaction. There’s someone on the other side. I see how they’re delivering the message, how they’re implicating. We’ll talk about it later in the context of the organization and what changes in companies. I don’t think it’s just judgment. I think there’s a little bit more than that. One of the reasons I went to the dark side of management early on in my career from being an engineer was Peter Drucker and this notion of the knowledge worker, which he later on reemphasized with the publishing of his book, which for me was seminal and defined a lot of my career in life, the post-capitalist society, which is this notion that information rich and information poor is going to be the key distinctiveness that will happen in the world. The two big camps, information rich, information poor, which links back to this invention of the term knowledge worker, that knowledge is going to be key in some ways. I think that’s what we’ve seen for the last decades. Again, I think judgment is not going anywhere, but I think it’s beyond judgment. There’s elements of humanity and involvement that won’t go away anytime soon, where human-in-the-loop are particularly critical. We’ll discuss later some scenarios, but for me, that’s my stick in the ground. I think human-in-the-loop is going to be critical for many decades to come.Bertrand Schmitt While we are talking about all of this, and we share some possible scenarios, there is always that question. This is moving so fast right now. If you think about AI 10 years ago, AI 5 years ago, AI with the launch of ChatGPT 3, and then AI the past 2 years, now we have agents that are running at scale. Things are moving very fast. I can tell you, me in 6 months, the change has been pretty dramatic in terms of what I can use AI for. There is always that question that whatever we are thinking about cannot just be connected to what we were able to do 6 months ago or even today, we have to think and project ourselves at least in the next 6–12 months. Of course, we can go beyond that, and we will do that with some future scenarios, but it’s a very fast-moving, and it’s not clear yet where are the limits.Nuno Gonçalves Pedro I think that’s a very fair point. Let me try to analyze things that I don’t think will change anytime soon for the next few years. Agreed with you that many things will change, and we’ll have a lot better tools, platforms out there. That will be difficult to predict what exactly won’t change. I think there’s elements of humanity, and some of them do relate to judgment, like having good or bad taste, having a view on it, on whether something looks good or bad. Obviously, all of this sometimes is subjective, but some of it may not be as subjective as people think it is. The elements of contextualization. I think a little bit going back to what we did at Chamaeleon ourselves, where we built this platform, Mantis, and the objective of building Mantis was not really to replace us, was that it was a core augmentation layer in some ways that we would use investment or investor judgment as humans in the loop to systematize pattern recognition and a variety of other things, but that Mantis would really elevate all that judgment, not just in terms of timing, us being more productive, but also in terms of the quality of the decisions we’re making. Think of it as a little bit like having our human judgment in the context of operating Chamaeleon at a higher altitude, where we are more aware of the things that are happening and how they actually happen. The ability to really get to the data pieces and then make decisions on top of that that generate the needed alpha in our case for investors. What I mean by this is I think there’s always going to be core elements of humanity that I do think are going to be difficult for the machines to replace. For example, the taste piece people are like, “I can figure out what’s the taste in the market.” Yeah, but that’s mainstream. That doesn’t identify what’s the next big thing, which normally doesn’t start from mainstream. It starts from something else. It could start from opinion leaders and influencers. It could start by someone having a different way of addressing a problem and having a solution that hasn’t been thought through. For example, elements of creativity, I think, in human judgment and in human operations is something that I feel the machine will still have difficulty to replace.Bertrand Schmitt Let’s not forget how today current algorithms are working by feeding them enormous quantity of data, actually as much data as we can find. Finding more data is becoming a limitation these days. What it means is that it’s very hard for AI to think beyond its training data. There is some level of logic that’s being added, but at the same time, take the launch of the iPhone. What was the opinion before launch? Is that no, it doesn’t make sense. Not enough battery life, no keyboard, no this, no that. If you just base your analysis on what’s written out there, what’s being sold out there, you would just say, “It’s going to fail.” AI might really follow that more generic advice and perspective because that’s what in the training data and that’s what they’re in volume. It’s, of course, raising a lot of questions of, how do you improve the quality of the training data? How do you separate the weed from the chaff? There are a lot of questions there, and obviously, it will get better over time. But it’s still a critical part of how it’s working today. It won’t be that easy to change. I really like your point regarding Mantis, and I will say in general, platforms that you build with AI or leveraging AI capacity. Because when we say knowledge production is going to disappear, but we’ll keep judgment, it will be a different type of judgment because the quantity and quality of knowledge we will have in front of us to build our judgment will be very different. If suddenly we have for free the work of 10 interns or 5 junior analysts or whatever, and you can run that on nearly anything you do in life or at work, it’s completely dramatic. Your judgment was not used to be exercised so often because often you were missing quality data to have a judgment. Before it was a lot of finger in the wind and trying to smell something, but you didn’t have enough to make a serious analysis. Except if you are working as a strategy consultant, as you used to do, Nuno. That part is actually quite interesting. That the judgment itself will be exercised much more often and hopefully on the base of much more in-depth analysis for a lot of things. We will work very differently.Nuno Gonçalves Pedro We will go in-depth, faster and more fact-based, more data-based along the way. The question some of you might have right now is, is there some judgment that’s going to go away? Is there some judgment? We seem to be defining that there’s this organization, we’ll talk about it later, that goes from doers more into deciders. I think there’s some nuances to that, so I’ll just hit pause on that. In terms of judgment, obviously, there’s judgment that has been hidden over the years under the pretense of being wisdom, but it’s actually not wisdom. It’s just repetitive tasking, and it’s rules-based for the most. There’s a lot of judgment done, in particular in the white-collar space, that you could say it’s just reps. People have been doing it all along like that, and so therefore to say, “I’ve done it before like this, so I’ll do it the same way.” There’s actually no best in class, no analysis, no nothing. It’s just, “I’ve done it like that before.” I think that type of judgment will disappear because, again, algorithms will be as good, if not much better at that. They’ll be better at figuring out, actually, this would be the better way to do this. That’s how you play it forward. Then the question is, if there are fundamental, wise people in the organization, people that can really take that more complex elements of judgment, how do you go from the world we have today, which is a world of apprenticeship, where people come out of college, they go and work, and they learn their way, and therefore, hopefully over time, some of them, not all of them, we know that, but some of them will develop that wisdom to be great decision makers 15, 20 years down the road? How do we do that in a world that now is saying, “I don’t need people out of college because I can do it myself, and I can do individual contributor, and I can have agents doing the work that would require some manifestation of management in the middle.” Basically, “I don’t need this stuff. I don’t need you.” It’s a little bit the story we’re in. How do you create then this apprenticeship? How do we create then wisdom? My two cents on that is that wisdom, because of what we were just discussing and what, for example, myself and Bertrand was just saying, because of more often interactions with more data-stressed information and insights, what will happen is people will get better through their own reps in whatever form they’re doing, in day-to-day life, in internships, et cetera. In some ways, that will create the accelerated growth. It’s a little bit the interactions with agents and the interactions with our beloved AI algorithms that will create that growth over time and maybe not as much with other people. That still leaves the question around social interactions, but that’s probably the way this gets sorted. Apprenticeship gets sorted through the machine and the human having more interactions in effect.Bertrand Schmitt I agree with you because when we talk about apprenticeship, in some ways a lot of time was wasted on stuff that were not that important. But in a way, that was the price you had to pay in order to be there when people make the big decision to try to get some wisdom from that one hour of interactions that’s really useful and make a difference out of your full week. But the rest of your full week was just basic stuff that you had to do like a machine in a way. Why not let a machine do that? That, for me, is a big question. You could argue there is a transition period where it could be hard. For instance, if you can work hand in hand with AI smartly while you are doing your 4, 5 years of universities, you could graduate with a very different knowledge, perspective, judgment, skill set than anyone who graduated 5 years ago. I think that part will require a question around, “How do you change education?” You see what I mean? If you keep education the same way, expecting that the output is someone that should go now into 5 years of apprenticeship, that’s not going to work because companies will be, “No apprenticeship anymore.” On the contrary, you have to come much more knowledgeable and ready to use the tools. The tools are so efficient that the bar pretty high. You need to come already very well-grounded. If the education is not doing their job, that will be trouble. That part for me, I think is often forgotten. In some ways, the new-found importance of universities as a place to, and not just universities, the trade to really deliver people who are ready for the workforce. If on the business side, the expectation can change, of course, you have to change the education on the other side. My worry probably right now is that it doesn’t look like universities are in touch with what businesses are looking for, businesses are working on. Of course, that’s very worrisome because the cost of university has increased very significantly. It’s not clear quality of education has improved at all. If anything, it could be the opposite. It’s pretty scary. Of course, it’s going to raise a lot of questions. How much is education worth in that type of situation? Maybe another point because we talk a lot about apprenticeship, how this stuff was useful, but at the same time, if we go back in time, not long ago in the ’50s, if you wanted to be a developer, for instance, ’50s, ’60s, the job was very different. There was barely any programmation language out there. You had to use punch cards. Your time truly spent doing the coding was very limited. Once you had your stuff working, then, the debugging was a total nightmare. My point is that no one is looking back to that time saying, “You know what? It was great. It was a great way to learn and to do an apprenticeship for 5 years. To do that crappy job of punching cards for the boss.” There was little value in this. Guess what? Everyone is happy it’s not being done anymore by anyone. I think we also have to see what AI is bringing in a similar way is that everyone’s job is going to become quite different. There are a lot of big parts of the job who are not going to look back with fondness. Just looking back as, “Wow, that was very machine-like type of job. I’m glad I’m done with it.” People will want to jump directly to the next step. You don’t need to go to the punch card phase to be able to be a good developer for the past 40 years. I guess it will be the same with AI.Nuno Gonçalves Pedro I think so. The difficulty we have as humans is to also visualize dramatically different scenarios and landscapes, professionally. It’s difficult for us to anticipate what are the jobs of the future. Jobs have changed a lot in the last few decades, not even the last century. What people do, the migration initially from the agricultural society to then the industrial society to then the services society, and in some ways, the shift within the services industry, and now we’re seeing another shift, so we can’t really anticipate what those jobs look like. Back to your point on education, because I think that’s a very important point. If you’re right now an undergraduate student or a postgraduate student, for that matter, and you’re not figuring out your own mechanisms of learning outside of your syllabus, outside of what your professors are telling you, et cetera, you’re going to face very difficult times. If you’re not right now using all these AI tools proficiently, all these cycles of vibe coding, co-working, et cetera, with agents in the mix, you’re going to have a really tough time. If you’re not at this point in time as proficient as someone like myself or Bertrand, and given that we’re nerds, we’re relatively proficient with a lot of these tools that are out there. On top of it, some of us have our own platforms in-house. If you’re not as proficient as we are with those tools, you’re going to have a very difficult time because then people like us won’t need you. I think that’s the sad truth. It’s like at some point, if you’re not needed, you’re not needed. Then again, you may find something else that’s more interesting for you to do. Start your own company, go join a new exciting job doing whatever it is that you need to do next, et cetera. But again, I think the bar is very high. If you’re in college right now, again, undergrad, postgraduate, this is the time of transition. This is the worst time. It’s not the best time, it’s the worst time. Because education and all these institutions haven’t adapted to it yet. You need to adapt. You need to adapt. You need to adapt. If you don’t, you’re going to pay for it, not just in the loans you need to repay, but also in terms of actually having difficulty finding your career path in those first few critical years.Bertrand Schmitt You need to be especially proactive when you’re facing this type of period where businesses are adapting as fast as they can because they all know it’s going to be survival of the fittest very quickly. Universities typically are working on a very different pace, and it’s pretty guaranteed they are not going to have adapted as fast as businesses. In time of big dramatic change, it will be trouble. It will be trouble. Yes, you will have not fun. Not saying it was part of the deal when you sign up for that loan and decided to go for university. But that’s life. There has been issues before. It’s not the first time. You have to do something about it. You talk about your perspective about, “Hey, why do we need you if you are not already fluent and very efficient with these tools and stuff?” The truth, in some ways, it’s even worse than that. Each time we spend with someone who is not efficient with all of this is less time we spend with the tools that are already providing magic for us.Nuno Gonçalves Pedro Exactly.Bertrand Schmitt It’s a very big choice of, “Hey, do I spend more time training this person?” Do I just… there is an opportunity cost. Or, do I spend more time staying at light speed? Why do I slow down to do something else in the hope that maybe I will get to return versus the light speed I’m already on? It’s a lot of tension. Again, it’s certainly new. But if we want to look back, I think you talk about the switch from agriculture and society, industrial society, and now the service industry. The reality is that, yes, we have made dramatic changes in the past before. 140 years ago, we were 90% agricultural society in Europe, in the US, 90% of us. Today, it’s what? 2%. So my point is that that’s a normal evolution. There is no progress without change. Sometimes the rate of change is soft, and sometimes you have a step function. Now it’s a step function, and it’s also a pretty fast step function. Before, it could take decades to get new stuff being put in place, to have electricity come up, this or that. Now we see that the rate of investment in AI is insane, way beyond anything we have seen before. Two, in a way, a lot of the architecture behind the scene was already there to support an even faster transition. What’s new might be the pace of the transition, how unnatural it might look. But at the same time, if you put yourself in the shoes of someone who lived 150 years ago, I mean, this was also a dramatic change for them. From horses to cars to planes to rockets, pretty big change, maybe even bigger change.Nuno Gonçalves Pedro Maybe the silver lining, just to bookend this section, is one, there will be new roles. There are a lot of things we can’t anticipate. There will be new roles, there will be new jobs being created, and new things that we can’t really quite grasp yet. The second part is that the rules are changing, and they’re changing, I would say, in general, for the better. If you are a decision-maker or an organization, and you still have your job, you’re probably making more important decisions with more data, with more tooling around you, with less red tape, hopefully over time. I know that will not hold true for all the big corporations out there that are listening to us, but it is starting to happen. Things are making an impact on how decision-making is made. There’s less and less red tape along the way in certain organizations. There are more and more fact-based discussions happening as we move along. The silver lining is better jobs, more jobs, different jobs in the future, hopefully as well. Secondly, the second part of the silver line is that the jobs that exist today, hopefully, will be more interesting, certainly on the knowledge space and on this judgment space that we’re now introducing as part of this episode. The AI-Native Company: Org, Hiring, Culture Switching gears, maybe to how does that shift? How does the company of the future look like? How does an AI native company look like? I feel there are a lot of discussions on, “Oh, you only need one person to run everything.” Let’s not go to that level. We’ve had a couple of episodes where we focused on AI as your co-founder and a couple of other elements that you guys can go back to. Let’s focus on a more evolutionary view of what’s happening to organizations, and maybe start with the org structure. In general, we should see more flat organizations where mid-level managers have to justify their pay in some ways because middle management are routers. They are normally routing tasks. It’s sometimes aggregating it, synthesizing it, and pulling it back up. Guess what? AI and agents in general are very good at that. The synthesis piece, et cetera, is not as well needed. One could say there are several elements of middle management that are valuable, like the coaching of people, the creation of apprentices, and the accountability that comes with some of middle management. But lo and behold, most of middle management is seen as a little bit of a thin line that doesn’t need to necessarily exist. I feel we’re moving into a world of smaller teams, more senior teams, where there’s more judgment at the top, where you’ll have people that both do a mix of what we used to call management in its new form, but also a lot of individual contribution. If you’re not used to that, if you’re not used anymore to be an individual in the future, again, and if you’re a very senior in an organization, maybe this is the right time to either reinvent yourself, find some other job that doesn’t require as much of that, which we’ll have plenty of those jobs for the next few decades, or maybe retire. I’ve actually, shockingly enough, seen people who have said, “You know what? This thing is changing too fast, too dramatically. My industry is changing quite aggressively right now. I’m about to retire in a couple of years. I’m just going to retire now.” I’ve literally met two people who have done that. Again, there’s nothing wrong about it. I think we’re, again, going through a step function and a huge shift, but figuring out where you fit in this new model of organizations, more senior at the top, smaller teams, more of a mix of individual contribution with management than ever was done before.Bertrand Schmitt I agree with you. In some ways, I’m not surprised that some people might say, “You know what? It’s now time to retire.” I feel a bit sad, maybe because it means you don’t like to keep reinventing yourself and changing your habits and thinking about new stuff. You were a creature of habits, I would say, if that’s your conclusion. But everyone is entitled to their own opinion, obviously, and a way of life. I guess that’s what happened, again, at regular times in the past in terms of big change. What I can see is that the rise of, you can call it the full-stack individual, someone who will have multiple roles inside the team. Before, you had to really separate the role. Especially in the US, there is such a clear separation between every role you can have in a company. Let’s take a tech company. You will have people doing design, people doing different types of designs, people doing front-end development, back-end development, and operations. You see step-by-step hyper-specialization. I have seen that, and it’s true that the level of complexity you had to deal with at some point requires some level of hyper-specialization because it will take you 6, 12 months in order to be really, really strong on a specific topic, a specific language. God forbid, trying to go deep into something that you had no real experience into. But I feel with AI, it’s a big change, actually. It’s the opportunity to go beyond that. It’s the opportunity to do more, to touch more. You can combine designing and shipping code, product managing and shipping code, being an analyst and deploying. Of course, we have to think how it works because putting a marketer shipping code to production, maybe that will get you into trouble. But I think that there must be some change. We see it changing dramatically, how fast we can get into something, something different from what we are used to. I think it would be crazy not to take that opportunity to dramatically change the scope of many positions and put an end to that hyper-specialization. I think for me, in some ways, hyper-specialization was bad. There is only so much you want to be a specialist in because a lot of things, a lot of opportunities are actually coming from the mixing of many different ideas, many different perspectives, and you lose if you go to hyper-specialization.Nuno Gonçalves Pedro I don’t think the age that is coming is the age of the generalist. I think it’s going to be the age of the multispecialist. We’re going to go into an age of multispecialization, which is a little bit, we’ve mentioned it as well in the past, what Amazon defines as an athlete or T-shaped or pie-shaped people, people that have on top an amazing ability to do general management, strategy, managing teams, et cetera, then have spikes. Spikes into business development, corporate development, product management, whatever it is. With AI and with agents, the development of those spikes, as we’ve been discussing in this episode, will actually be easier. It’s almost like a given. If you want to go deeper and deeper into a certain area, you can go much faster. I think that level of multispecialization is going to be really cool to observe. I’m not sure we’ve had an age of multispecialization over the years. Maybe people would point out, well, the Da Vinci example, people that are great across very different areas. Maybe that’s an example of multispecialization. But honestly, from my perspective, this is going to be an exciting time because of that, because you’ll have people who, instead of being just focused on this area of sales, and I only do that, they can actually and should actually do a lot of other things. So the work, as we were talking before, can be more interesting. More demanding as well, because the judgments you need to make are more complex. The context you need to actually gain needs to be gained much faster. At a level of magnitude, you haven’t been able to do it before. Talk about information overload. But actually, ultimately, the roles can be a lot more interesting, a lot more exciting, because I can jump around. If I’m an investor, in this case, we have two investors on this conversation. But if I’m an investor, one of the things that we start looking at is actually not just looking at a startup as, is this startup doing something in AI or not? Is it AI-enabled or not? Is it an AI platform or not? But actually, more fundamentally, is this an AI native startup? Meaning, organizationally, culturally, is this the company that’s already in the AI age? How is the team working? How are they defining things? It’s not just that they only have two or three people. It’s like, what are those two or three people doing? How are they doing it? What cadence are they doing it on? What tools are they using? How are they making decisions? I feel we’re still actually relatively early on that track. It’s very interesting because we’ve had all these companies raising mega rounds. First round out, we invested in one of them, but there have been many frontier labs out there raising a ton of money. But a lot of them don’t have a fundamentally different way of doing business. Of organizing themselves, of how they do the day-to-day. Although they’re working on cutting-edge stuff, with very notable exceptions, they’re actually not using it themselves. They’re not actually shifting how they do stuff themselves.Bertrand Schmitt For me, that’s very interesting because in the past, I used to be quite conservative on how you manage and run a company in the sense that if you’re already in tech, if you are already on the cutting edge of what technology can deliver, and this and that, don’t waste time trying to invent a new org structure. Just focus on delivering something great, amazing, and be great at technologies. That’s already your huge differentiator. At the time, there was no real reason to innovate on the team organization. I have seen so many teams that tried to innovate, and it was just catastrophic because there was not much to innovate on, because we had decades of optimization that we could leverage. There was no reason to invent. But here it’s very different. There is a dramatic shift in how you can organize differently a company. I don’t think there are any blueprints yet on what’s the best way to do it because it’s too new. But at the same time, I would feel very bad to invest or support a company that first is not focused on AI or AI-enabled, but at the same time is not trying to innovate on the team itself. Because if you don’t do that, you’re going to get killed by someone who is going to innovate better than you on not just the product, but on the org as well.Nuno Gonçalves Pedro Indeed. The shifts are pretty substantial. If you look, for example, just at hiring, what do you hire for? Certainly, there’s this element of the multispecialized orchestrator, which normally will be someone with quite a lot of wisdom and expertise. It doesn’t necessarily mean someone who’s old, but someone who has the ability to work with all the AI tooling and platforms out there and be an orchestrator of agents. Why do they make judgments, make decisions, move stuff forward really, really, really quickly? Again, those jobs are going to be the best jobs. The second part, I think that is very interesting, around hiring, is you’re going to skew towards the elements that are potentially either very aligned with the use of AI tooling and platform, AI expertise, or being AI native, or someone who’s used to using AI. That’s one side of the fence. On the other side, you’re going to actually be optimizing to hire people that have the characteristics that will be difficult for AI to replace immediately, like taste and the notion of fundamental accountability and notion of implications, the notion of how you affect change in organizations, how you affect change in individuals, the elements of coaching, and beyond coaching. You’ll be optimizing for those kinds of hires as well. Then, last but not least, for me, I feel that there is a momentum already happening. I think it will happen even more, which is the tendency to under-hire rather than over-hire. The moment of the good old days of blitz scaling, “Oh, let me go and hire 300 people to scale my go-to-market and just land grab market.” Now, that’s not how it’s going to work. People are going to try and first get the efficiencies in-house with top talent and see if there’s, at the end, the need to hire more people or not, rather than the other way around. I think the issue here is a little bit of what we alluded to before in this episode. There is a tax on individuals. If you hire more people, you’ll have to manage people, you’ll have to work with them, et cetera. If I don’t need to, I might as well work with the agents that the tools and platforms that I use give me access to. Because that’s a world that’s much more efficient, right?Bertrand Schmitt I’m in total agreement with you on this. It’s definitely raising way more questions than before because, again, on one side, you have the product, the technology used to build products that are completely different. At the same time, all of this is also enabling new ways to design organizations and to scale differently, especially in a world where, as we have seen in 3, 6, and 12 months, stuff that you thought were impossible are suddenly becoming possible. So you’re, “Hey, I’m going to scale and burn a shitload of money for 6 months before I know if there is any return.” Versus, “You know what? Maybe I just wait 6 months. The AI has improved enough so that we don’t need this new team. We don’t need these people to do stuff.” Because actually, if you just wait 6 months, we will have stuff coming for free from either new AI models or new AI tools or this or that. If you remember, we used to say that in mobile, things were going three times as fast as on the web in terms of pace of innovation and speed of development and stuff. I mean, with AI, it’s 5X mobile.Nuno Gonçalves Pedro Maybe even more. Yes, well.Bertrand Schmitt Maybe even more, maybe 10X. Every assumption around blitz scaling or scaling in general was based on past assumptions. It’s not based on how is the industry evolving today. Might make more sense for you to really grow your agents and spend more money on more tokens. I remember, of course, Jensen is selling his business interest, but he was saying, “Hey, for each one of my 450K engineers, he better spend 250K in tokens a year.” I’m not saying it’s the right way to say it, but I think there is some truth in it, and that would be something to think about. Have we maxed out the token usage per employee? I’m not talking in a stupid way because token maxing and wasting money has no value and is as stupid as it gets. But if you are truly getting a return on these tokens, can you use more? Can you generate more? Can you create more loops so that one engineer manages not just 10 agents, but 50 agents, but 200 agents? I think that’s the big question. We’re trying to add more people. More people means more management, more issues, more this, more that. That would be a fair question. Another piece of the puzzle is how do you build in a way your… I don’t know if it’s a digital twin, but more like the digital version of your companies represented by agents. How do you make sure that everything you do as a business is truly captured, is truly leveraged so that your agents are getting better and better? Not just because the model gets better, but because you are putting more data into it, because it has more opportunity to learn, and as a result, gets better at your specific business.Nuno Gonçalves Pedro The next big thing is culture. How does culture change? I think the biggest shift that I see is, why would you do meetings all the time?Bertrand Schmitt Yes.Nuno Gonçalves Pedro At least at Chamaeleon, we have a very small team, just by the way. We have a very small team at Chamaeleon. We’ve reduced by way more than 50% the time we spend on meetings between each other across the board, one-on-ones, partner meetings, et cetera. I think we’re really pushing to be more and more asynchronous. There’s stuff you can process via message. I was just asking one of my colleagues, “Can you just send me that prompt for that so I can just do that on CoWork?” Or “Can I just go on Mantis and do this? Can you tell me the cycle?” Or vice versa. Basically, it’s a little bit like you’re just going to do it. I don’t need to meet. I don’t need to meet all the time. There are some things where we still need to meet and interact, and we need to brainstorm at times, and we need to go to a different level of abstraction on the top end. Then on the lower end, there might be things that are a little bit more specific and governance-related and operational-related that we need to agree on that are more sticky. But otherwise, the culture is going to be biased towards build. “Go and do it,” rather than, “Let’s do a meeting.”Bertrand Schmitt Yes.Nuno Gonçalves Pedro Async is the thing. I’m more and more like we have a couple of interns this summer. “Can we async this?” They’re like, “What does that mean?” “Can we make this interaction asynchronous?” Because synchronous interactions for me are very expensive. Can you send me something that I can process, and then I can send it back to you? We don’t waste time on you giving me context and whatever. Then I’m not ready quite yet because I need to process it. Maybe I’m in between two meetings that I’m actually thinking about other things in my mind.” Again, I feel that shifts how stuff is done. One, build rather than meeting. Two, asynchronous versus synchronous. In some way, millennials had it right when they shifted a lot to messaging and stuff like that. Let’s do more asynchronous rather than synchronous, those two elements from just an operating model of the company are significant. Maybe this is a good time for me just to put one parenthesis because there’s this thing that’s bugging me as we’re talking here. Everyone who is listening to us at this point in time might be saying, “Cool, but I work for this large organization. We’re just now…” Everything we’re saying here is contextualized by time. We’re giving you extreme situations. We’re looking into the future. Some companies that we’re talking about might be doing this already as we speak. Some of them might be in the process of doing this and might in the next couple of months be doing it like we are describing it here. Some of them might take years to get there. Then again, some of the companies that might take years might actually be destroyed in between or meanwhile, and be disrupted. Some of them might not because they’re in very legacy businesses, and it’s fine, and it’s okay. Again, don’t take everything that Bertrand and I are saying today as this is gospel, and it’s going to happen tomorrow, and why the hell are we not doing it? We think that aspirationally, this is where you should be moving to as an organization, whatever size you’re at. Speed will matter, as we discussed before, but not everyone, obviously, is going to move as fast as we’re describing it here.Bertrand Schmitt Yes. Me, for instance, take inspiration often with what some of the AI labs, frontier AI labs, are doing, the way they are working, especially in OpenAI and Anthropic. They are clearly at the top of the spear in terms of what is it that you can do because they have access to models we don’t have access to, because they have unlimited tokens they can use for tasks. They hire people who are, of course, 100% on AI. They are the best example of what is achievable if you have the top minds, if you have the latest models, if you have unlimited tokens. From there, you can take that for our needs and for our situation, and others in industries that are not as advanced. Definitely, you have some time. But as you say, things are moving fast, things are changing. Wall Street is going to expect better returns because when we discuss all of this, the conclusion is that you should be able to do more with less. That’s as real as it gets at some point. By the way, that’s what you see. You see better performance, a better business performance right now. So even if you might not get disrupted, you’d better start there. For some, it might take more time, and they might still be fine.Nuno Gonçalves Pedro Maybe to bookend this section, clearly what we’re saying is organizations are going to change. Their MOs are going to change, the structures are going to change. There are elements of what we discussed before in terms of judgment that are fundamental to this. The ability that in some ways, one would say a lot of the technique of getting solutions out there, even in brainstorming or problem-solving, is going to get democratized. The algorithms are able to do that. On the other hand, having points of view and having wisdom is not necessarily democratized, necessarily by the machines. It can be facilitated, it can be more productive in achieving that level of wisdom, but wisdom still will matter at the end of the day. We’re not saying that’s out of the question. Actually, that’s going to be the asset. People who have fundamental wisdom that can come to the table and frame things. We see this even today in prompt engineering, on just creating prompts. The better your prompt is, the better the outcome is going to be, the result that you get from the algorithms. That’s not going to change, in my opinion, anytime soon. That UI interaction piece is not going to change anytime soon. Again, if you’re an organization thinking through organizational structure, culture, if you’re thinking through hiring, these are some of the elements that we think will give you an opportunity, but I would actually go one step further. On the positive side, I would say, they give you arbitrage. If you’re able to move faster than your competitors and really adapt your org faster, you’ll reap the benefits faster as well. That’s what many still say and relate to as the word innovation. That’s how innovation gets accelerated. I think there’s a huge opportunity right now for arbitrage. If you move fast, experiment, experiment on new org structures, experiment with talent, you’ll know that some of them will work well, some of them will fail miserably, so you can’t experiment on literally everything. On the other side, I think the doomsday scenario is if you don’t, if you’re on the other side and your competitor is outpacing you on trying these different organizational models, structure, hiring models, and operating models, they’ll potentially just disrupt you. They’ll do stuff that you thought you had the moat on, and lo and behold, you don’t anymore. Sometimes it comes just from org, just from injection of people with a different MRO, different operating model.Bertrand Schmitt The Human Element: Are We Underestimating It? Maybe we can move to our next section about the human elements. Are we underestimating it or are we overestimating it? The three things that are a big part of the human elements, emotion, creativity, and synthesis. Is it just soft skills, replaceable part? On the contrary, is it the durable part now that we have automated intelligence?Nuno Gonçalves Pedro I’ll start with emotion first because I think it’s probably the easiest of all the ones you’ve mentioned. Emotion is key. Many of you listening to us will know this. The way you deliver a certain message, the emotion that you have when you deliver it, just in and of itself, this could be a sentence, it’s something verbal, et cetera. Makes a difference between the person or the people on the other side actually adopting it or actually just resisting it. Emotion is critical. It’s what runs the world. Everyone talks about a bunch of things, but emotion is a currency that is still naturally human. It will be, I feel, difficult for these AI tools and platforms to recreate it fully until there’s some literally very high-definition manifestation of them as avatars or some physical manifestation of them as robots and all that stuff. It will take a while for that emotion to be manifested. Emotion, I think, is still something that we as humans have as a moat, and it’s critical. As you mentioned before, I was a strategy management consultant at McKinsey, and getting people to action is actually 80% about the delivery, communication, the emotion that you surround the project itself, more than sometimes the truth. It’s great to have the truth and to have something that is similar to the truth in terms of analysis, but in some ways, that’s not what really moves change. Change is moved by, I would argue, a significant amount of emotion and alignment on emotions.Bertrand Schmitt You could argue that’s something that most politicians have perfectly understood. If you look at most campaigns these days, everything on emotions, maybe the tagline might be one word. It’s interesting when you see from that perspective that actually it’s very little on facts, very little on all of this, but more about emotion. You could argue it’s the same for businesses in the future? That’s a fair question. I think creativity is another one that’s quite important. At the same time, it’s not so easy because I must say I’m quite amazed when I’m looking for creativity from AI, either to generate the image, to generate video, to generate audio, or to generate text. AI can be pretty creative. I still think you need to control its creativity; you need to understand what’s good, what’s bad, what’s quality, but at the same time, I can see even in creative tasks, AI can be a very strong partner. I’m talking about any creative task, like invent a name for a product, let’s brainstorm the mission for the company. AI can actually be doing a pretty impressive job. That’s the type of job where you will hire experts, where you will use some of the best people in your team to help you for days. We say, “You can do quite a lot.” It’s an interesting one because I think there is some unique human creativity, and at the same time, AI can be pretty strong at creative task as well.Nuno Gonçalves Pedro I agree. In particular, if it represents benchmarking, if it represents repetition, if it represents seeing the world and then coming up with something that presents itself as creative, to be honest, it can actually outpace humans. If it’s like genuine light bulb moments of creativity, angles that haven’t been tried before, certainly not in the same way, I think humans still have the advantage. To your point, I agree. This is not a humans-win situation. On the previous one, on emotion, still, part of it is because, also on emotion, there are exchanges. You and I might be looking at each other, and from the facial expressions and the reactions, where you judge that for AI to get there, it’s going to take a long time. There’s going to be a lot of very complex algorithmic stuff put into that for AI to be able to create synthetic emotional behaviors, but creativity, I agree with you. There are a lot more nuances to it today, where AI does have significant advantages at the end of the day. Synthesis depends. Synthesis, I feel, if we’re talking about holding a bunch of messy assumptions, contextualized inputs with different layers of data adjacent to them and then trying to create and form one coherent, fully accountable point of view that you stake something on, like a decision, a company, a business unit, whatever, I think humans have the advantage. Part of it is the complexity of what we have today with generative, pre-trained transformers, today with GPTs, where the hallucination comes through, where it’s really more statistical analysis. Over time, maybe synthesis will be a forte for AI. Right now, I think we still have that ability to really be the ultimate decision-makers and judge-makers and have that wisdom put at the table to make those decisions. Honestly, models are very good on balancing both sides, so ended up, as we say in Portuguese, neither fish nor meat. It’s to balance both sides’ answers. That’s not helpful in most cases. When you’re in a difficult position where, for example, the future of a company, company is almost dying, what do you do? I’m not sure your AI algorithms that are going to give you a great solution. Because it will give you a median or average solution, which likely will lead you to a median or average outcome, which in this case would be failure. Again, on synthesis, there are some areas of advantage for human beings. If you are looking for clearly synthesized perspectives on certain elements that are maybe less edge-focused, they’re more than the normal part of the normal distribution, then probably AI agents are brilliant at that. All the tools we have today are pretty good at that, and I think they’ll just get better over time. That’s how I see synthesis.Bertrand Schmitt I think a lot of improvements will come with a better fine-tuning of agents to what’s special about your company. Because if you just take a general agent, there is only so much. It can understand your industry, your company, and your way of working. I think that part of making sure your agents are finely trained, finely tuned on your own business, so that they can give you a really well-calibrated feedback, will have a lot of importance.Nuno Gonçalves Pedro I think that’s absolutely spot on. Maybe to end it, what is definitely different about humanity? Definitely, emotion, as we discussed, some pieces of synthesis. Creativity, maybe the light bulb creativity, not the more repeatable creativity, the one that you can put and encapsulate into processes in some ways. There are elements of us being physical, which robots can’t still recreate. That’s definitely an advantage. The embodied, we’re embodied. That’s obviously a huge advantage. With that also comes advantages because we have to interpret each other, and we have to see the complexities in physicality that land to it. Is human and the human element categorical difference? If we’re having a more philosophical discussion around this, I think it is. I think it will be for at least the foreseeable future and maybe decades to come, even in whatever scenarios we’ll discuss, which is our next section, scenarios.Bertrand Schmitt I would say projecting beyond 10 years is always pretty hard on this because, again, some of the improvements we are talking about we can imagine based on how it has evolved, but at the same time, there will be disruptions in AI. Stuff that we take for granted in terms of weakness, especially, might not be there in a few years from now. Either because it has been solved through brute force or because the field will have made significant change and improvements and discoveries, making some of our points moot. If we talk about embodiment, obviously, robots are coming. How fast, how cheap? That will be a big question. Right now, they’re not very smart. They’re usually very specialized. The more we move to a more general form factor, humanoid form factor, the more I think it will change. Also, another piece of the puzzle is that we have the assumption of agents having trouble to convince humans and stuff. At some point, we keep assuming that humans in the loop. If we’re talking about agents convincing another agent, not having embodiment might be even more efficient. That will be another perspective. Going forward, we will have not just agents we control who are doing a job and scanning the job, but agents truly interacting with other agents. You have agents controlled by one person, one team in your company, working either together or maybe not confrontationally, but trying to think and having different perspectives with another agent, controlled by other teams. I don’t think we have seen much of that now. We have seen mostly agents that are controlled by one team doing one job in one direction. Not multiple teams agents working together, or against or in parallel with another team agent. I think we will see some interesting things coming out of that.Nuno Gonçalves Pedro Scenarios Switching to scenarios, we love our two-by-twos. We haven’t done one in a while. This time it’s a two by two. We have four scenarios. I think on one axis, we would have potentially the capabilities of AI. One side would be more incremental. The other side would be the extreme full AGI. I’ll define it in a bit so that we can at least have a little bit of a definitional view on what the AGI is. Then the other axis would be how gains are distributed, concentrated versus broad. Obviously, if they’re very concentrated, it’s more unequal. It only goes to a few companies, a few people, a few individuals. If it’s broad, it’s much more dispersed through society, et cetera. AGI, just to try to define it, the formal definition of it is that it’s a hypothetical AI that matches or exceeds human capabilities across virtually all cognitive and practical tasks. In some ways, AGI can learn, reason, and adapt to novel situations across any domain. Then there are several mutations on this, but there’s one notion, or rather, there are three notions that normally are across a lot of these definitions. One is generalization, ability to seamlessly transfer knowledge from one domain to another without needing retraining, which is a very impressive skill that we humans still seemingly have. Autonomy in agency, the capacity to operate independently, set goals, plan and execute complex tasks. I think AI is their issue with agents to a lot of that extent. Then, last but not least, human parity, performing economically valuable work at or above the level of a typical human knowledge worker. If you listen to one of our last episodes, you’ll realize that Bertrand and I have slightly different views on AGI, and if it’s already here or not. I think, definitionally, maybe we have slightly different views on what the definition actually is. For me, maybe AGI is a little bit more what some would call superintelligence and generalized superintelligence. Strict to census, Bertrand is more connecting to AGI as in its prime definition. It behaves as well or better than a human thing. Maybe that’s what’s leading us to differences on whether AGI has arrived or not.Bertrand Schmitt Personally, I will have a different scale where I will put AGI, as you just said, in some ways, relatively similar in performance to your average human being. On top of it, it’s able to touch different domains that most humans are not able to do. Usually, there is some level of specializations where in AI, it can be more generic. I will put ASI, Artificial Superintelligence, as clearly the step beyond. Something that, on any dimension you pick, it’s able to beat a human expert. From my perspective, I think we already discussed that, but we are at AGI already. We have AI that can do way better, not just way better, but at least as well as humans on many topics, sometimes better. Yes, there are some topics that are not for AI yet. Embodiment, for instance, to flock with your humanoid robot in 2026. For me, we are partially there or fully there in AGI. If we take the stricter definition, ASI, we are definitely not there, but my guess is that it’s moving quite fast. We might be there in a few years from now. I don’t think we are talking about multi-decades. It’s 5 years, maybe 10. Of course, there are questions because people will say, for instance, “Hey, how do you become truly super-intelligent when all your training is based on human data?” That’s not an easy one because how do you train on that? To be way better, not just a bit better, but way better. Maybe I’m going on a tangent, but some are looking at AI learning from AI, AI being taught from AI, AI fighting with AI, AI challenging AI. The same way we saw this AlphaGo moment where AI was not trained anymore, like in chess with human moves, but has been trained to play against itself. That’s when it reached superintelligence in Go. It reached superintelligence by playing against itself and basically letting go of that human baggage, if you want, and going to the next level. What I found interesting in that, actually, first, that’s what happened, but two, there was some analysis that the average level of Go players and the top players went up after AlphaGo because AlphaGo, in a way, opened doors that humans didn’t believe were open in front of them, or they didn’t see them. They didn’t see these doors, so they didn’t bother to open them. AI opened new doors, but interestingly enough, humans improved after that, thanks to AI. You see what I mean? It was an interesting, okay, that self-learning from AI was the way to go beyond the current level of human knowledge and human expertise, but at the same time, humans were able to follow up. It was not like suddenly humans are totally useless crap. They improved. Did they still beat AI? Maybe not, but it was definitely also helpful.Nuno Gonçalves Pedro Back to our scenarios. We’re going to take the definitional extreme just for argument’s sake for scenarios. We’re going to talk about maybe what you were saying, ASI rather than full AGI, but like ASI. Again, artificial superintelligence as the extreme on the one hand. Let me talk about maybe the first scenario that would come to mind. Maybe we can call it the plateau scenario. All of this was great, but it was all smoke and mirrors. They were great at some cognition stuff. They’re a great tool. At some point, they’re going to hit a wall. Hallucinations are never going to be a thing of the past. We can’t fully trust them on really hardcore stuff. We’ll gain productivity enhancements. We’ll keep gaining those productivity enhancements, but at some point in time, we really won’t reach ASI. We really will be stuck with what we have. It’s a little bit like we get the next big thing, the next big spreadsheet, the next big internet, but it’s not going to change the whole world beyond just productivity, enhancements, and amazing tools that we have available to us that makes us much better. In that scenario, the winners will continue being fast adopters, probably small and medium businesses, because there won’t be a push for maximum speed either, so they’ll catch up at some point. Then AI native companies will be better companies than other companies, but not necessarily overall disruptors across the board. It’s not necessarily a new species of companies. It’s just companies that are a little bit better at doing stuff, which we also saw during the internet phenomenon and that first big push forward and then bubble, where we had some companies that were fundamentally different on how they operated. It took us another couple of decades for companies to be more and more digitally native along the way. Basically interesting, but it’s boring. It’s like, cool, we got tools, we got promised the world. What are the implications? All these companies that are worth trillions and trillions of dollars are not worth trillions and trillions of dollars. Because at some point we’ll face competition, commoditization. It will just be tools and platforms. They will not unlock that next stage. Therefore, this will have been a bubble, and likely it would be a hard landing to that bubble. That’s the implication.Bertrand Schmitt I would just say that, yes, I agree with you, but I would just say overall, even if it stopped today in terms of quality improvement, speed or stuff, or it barely improves, I still think we will have 10 years of madness just to leverage everything that we have today.Nuno Gonçalves Pedro Understood, Bertrand. This is a scenario. I understand, but maybe we’re going to hit a wall, and we’re going to hit that wall next year, or we’re going to hit that wall in 2 years or whatever.Bertrand Schmitt Possibly. I’m just saying we still have 10 years of goodness from that big push in AI we experienced the past few years.Nuno Gonçalves Pedro Absolutely. Agreed, but it’s boring.Bertrand Schmitt It’s boring. It’s a plateau.Nuno Gonçalves Pedro It’s a plateau. The second one is more of something that we have AI, but humans in the loop are going to be critical along the way. The judgment work that we described earlier in the episode is going to be critical to everything that happens. It’s, I would call it the augmentation scenario. The AI will be a great augmentation tool for humans, but humans will never really quite stop being in the loop. Some of the gains that AI has are broadly distributed in society and in the startup, big corporation and small medium business world. Everyone will have access to them. We humans, are still very important. We have all these augmentation things, and AI is mostly benign. There will be a couple of issues, but honestly, at the end of the day, we’re just better. We’re better, faster, more data-driven, more factually current. We’re doing stuff faster, but humans

A inteligência artificial está evoluindo em uma velocidade impressionante. Mas será que estamos apenas vivendo mais uma revolução tecnológica... ou já entramos na chamada Singularidade Tecnológica?Neste episódio do Bunker X, recebemos @izzynobre Izzy Nobre para explicar o que realmente significa esse conceito e por que ele vem sendo discutido pelos maiores especialistas do mundo.Falamos sobre LLMs, AGI, Superinteligência e mostramos casos reais que surpreenderam até os próprios criadores das IAs, envolvendo empresas como OpenAI, Google DeepMind, Anthropic, Microsoft e Meta. Também discutimos o que essas inteligências já conseguem fazer hoje e quais são os desafios para mantê-las sob controle.No final, entramos no campo das hipóteses: estamos caminhando para um futuro como Os Jetsons, para um cenário digno de Matrix e O Exterminador do Futuro, ou a Singularidade já começou sem que a gente tenha percebido?#InteligenciaArtificial #IA #AGI #SingularidadeTecnologica------------------Este programa foi um oferecimento de:INSIDER. Garanta descontos incríveis usando o cupom BUNKERX em: https://www.insiderstore.com.br/BunkerX#InsiderStore------------------

Hey, it's Alex (yeah, I'm finally back from my vacation!) What a freaking week to come back to! Just after our last episode was published, Anthropic releases Opus 5, Jensen joins X and drops the “Open Weights & AI Leadership” open letter, Kimi K3 is released the following Monday beating expectations, and then the AI hack (OpenAI model breaking sandbox and infiltrating HuggingFace) is on everyone's mind, another Open Letter, this time from over 1K employees inside the frontier AI companies all talk about pacing the pace of frontier AI development. We played with Opus 5 and Kimi K3, and had the great pleasure to chat with friends of the pod Elie Bakouch (Prime Intellect) and Philip Kiely (BaseTen) about this important open weights release, then covered our general thoughts on Opus 5, and made order of all the different open letters that came out this week. Finally we chatted with Max from Pangram about the next version of AI writing detection (their biggest yet) and finished with Zuckerbergs (also on X! what's going on with everyone joining X) op-ed on the vision of personal superintelligence for everyone. Let's dive into this (as always, all the links and sources at the end, please don't forget to sub to our podcast on your favorite podcast app!) Open Weights AIKimi K3 the king of open weights - 2.8T chonker MoE near frontier model (X, HF, Blog, Tech report)This has got to be the biggest news of this week, and maybe the open weights AI news since GLM 5.2. MoonShot came back with Kimi K3, and we haven't seen any models quite this large in the open. Even Grok 4.5 is around 1.5T, this model is nearly 2x the size. Coming in at close to 3T parameters (and 2.5terabytes of weights at MXFP4 format), this model comes in very close to frontier! This was such an important release that I invited 2 friends of the pod, Elie Bakouch (prev HuggingFace, now Prime Intellect) and Philip Kiely (Author of Inference Engineering book, BaseTen) to dive deep into what makes this special! Elie's take, from reading the tech report, there's no single secret sauce, it's a combination of already available in the open techniques. Like KDA (Kimi Delta Attention) that has been out for a while, attention residuals, NVIDIA's latent MoEs. The highlight for Elie was the scaling work they did that reported a 2.5x scaling efficiency over Kimi K2.5 (2.5 performance at the same compute)! They also skipped RoPE entirely in favor of NoPE (the report calls it No Positional Encoding) for long context.Serving 1.4TB on eight GB300s (Baseten blog)Philip's team at Baseten was a day-zero provider (we're still working on bringing this model to CW Inference, stay tuned!) so I invited him to tell us behind the scenes of hosting this beast. Philip said that just loading the weights takes about 1.5TB!! of VRAM, and that's before the KV cache allocation + 1M token windows, so they're serving it on 8 GB300s where NVL72 . Baseten worked with the vLLM and SGLang teams on kernels and he also said they contributed patches back upstream! The model was trained with MXFP4, which, unlike Nvidia's own NVFP4 is a more standard format per Philip. I enjoyed his deep dive analysis into the differences, but because of this and because they trained the model with quantization awareness, it's “only” 1.5TB vs the would-be 5-6 TB if that this model in FP16 would demand. One of the more favorite nerd snipes moments, Philip pointed out that his colleague discovered that with over 99% of the usage being cached (think harnesses that send millions of the same cached tokens back and forth), tokenization actually starts to become a bottleneck. So they released a custom “basetenkenizer” that reduces the latency to serve the first token significantly! Great job!The harness in question is very importantOne important callout with 2 evidence pieces - the way you inference this model really matters. Kimi trained K3 with preserving thinking history, so when your harness uses it, it must send back the full thinking and tool use into the API to get the best next response. If your harness strips that out, you're not getting the most intelligence out of Kimi (shoutout to Niels from HF team for pointing this out). Additionally, the Composio folks, tested K3 on 3 harnesses, Kimi Code, Hermes and Claude Code. The difference in outcome was negligible, but the different in cost and number of tokens is definitely surprising! Claude Code (as a harness only) took 9x more Kimi tokens to get the same responses! This is also why Kimi Vendor Verified exists, their own held back benchmark of how well model providers serve Kimi across different quantization, tokenizer and KV cache settings. Benchmarks and the license! Ok let's start with the ugly... this isn't MIT, not remotely. This model is suspiciously served by all providers with exactly the same price (check OpenRouter) and requires inference companies to sign a contract with Kimi (I've no internal knowledge of this except that CW folks are working on it). Not something I particularly like, but hey... we're still advancing the frontier here! Speaking of frontier, this model approaches the frontier very closely. On DeepSWE, K3 sits just behind Fable 5 and GPT-5.6 Sol at 67%, beating GPT-5.5 & Opus 4.8. On Terminal-Bench 2.1 it takes second place behind GPT 5.6 Sol! It's 4th overall on Agentic Arena, with frontend design being genuinely good across the board - 1st on Design Arena

What if the thoughts running through your mind are not there to be controlled, suppressed or eliminated, but simply observed?In this series, I select my favourite and most insightful moments from previous episodes of the podcast.Today, my guest Banya Lim, a multi-generational healer and energy coach, shares a powerful teaching on the connection between thoughts, emotions and the body. She explains why so much emotional suffering comes from identifying with thoughts, and how allowing them to pass without resistance can restore a natural sense of balance within.Press play to learn how to relate differently to your thoughts, release emotional tension, and reconnect with a calmer state of being.˚VALUABLE RESOURCES:Listen to the full conversation with Banya Lim in episode #518:https://personaldevelopmentmasterypodcast.com/518˚Coaching with Agi: https://personaldevelopmentmasterypodcast.com/mentor˚

En este episodio de Creadores Podcast, nos sentamos con Juan P. Navarro, fundador y experto en desarrollo de software e inteligencia artificial, para analizar cómo la IA está transformando radicalmente los negocios, el mercado laboral y la toma de decisiones estratégicas. Juan P. explica por qué las agencias tradicionales y roles como los clippers o editores tienen los días contados si no aprenden a integrar herramientas avanzadas, y cómo la verdadera oportunidad radica en la orquestación de agentes de IA y automatizaciones profundas. Descubre la diferencia crítica entre usar modelos de lenguaje como ChatGPT o Claude a nivel personal —lo cual puede perjudicar tu plasticidad cerebral— y utilizarlos a nivel profesional para ganar velocidad, productividad y escala sin precedentes. Shownotes 00:00 - Por qué ChatGPT puede ser peligroso para tu cabeza01:47 - Cómo ver la IA como una oportunidad de negocio real04:49 - Por qué las agencias de marketing tradicionales van a quebrar05:31 - Qué es un LLM (Large Language Model) explicándole a un principiante08:10 - La verdad sobre las diferencias entre modelos de IA y el costo de tokens10:17 - Cómo funcionan los tokens, inputs y outputs en ChatGPT14:30 - Qué es AGI vs. ASI y por qué la superinteligencia artificial nunca llegará17:19 - El caso Anthropic (Fable 5 / Mythos 5) y los riesgos de ciberseguridad26:47 - El baneo de la IA en China y la venta cancelada de Manus AI a Meta28:42 - Los 5 errores que provocan despidos por IA y cómo evitarlos33:18 - Diferencias clave entre usar IA como consumidor vs. usar IA en negocios36:05 - Por qué ChatGPT daña tu plasticidad cerebral a nivel personal42:22 - La verdad sobre la psicopancia en IA: por qué ChatGPT te miente44:54 - Cómo usar la Inteligencia Artificial a nivel profesional para ganar velocidad47:15 - Cómo hacer Prompt Engineering avanzado con contexto y estructuras ganadoras56:53 - Tipos de memoria en IA: memoria nativa vs. bases de datos vectoriales (Supabase)1:00:10 - Cómo crear workflows agénticos conectando N8N, IA y YouTube1:02:30 - Qué es un Agente de IA y cómo crear empleados digitales autónomos1:08:10 - Necesitas aprender a programar para crear Agentes de IA?1:11:45 - Cuándo es una mala idea automatizar procesos dentro de tu empresa1:13:32 - Cómo un CEO o Founder debe usar su propio Agente Ejecutivo en Telegram1:21:57 - Cómo usar Inteligencia Artificial en el proceso de ventas sin perder cierres1:26:45 - Qué hacer cuando la IA reemplaza el trabajo de un Marketer o Editor1:32:07 - La verdad sobre el contenido generado con IA para redes sociales1:36:05 - El futuro del trabajo: oportunidades reales vs. miedo por venta de cursosSi te gustó este episodio, te recomendamos ver:- https://youtu.be/Y_V30K2JL9Q- Recibe 5% de descuento en tu suscripción de los mejores suplementos utilizando el código CREADORES en https://belevels.com/- Recibe acceso gratuito a mi lista de los 100 libros que transformarán tu vida aquí: https://www.creadores.co/newsletter- Únete a la lista de espera para nuestro proximo evento presencial en CDMX: https://creadores.co/eventos/- Invierte en bienes raíces en EE. UU. con nosotros en Creadores Capital y genera retornos promedio del 20% anuales. Aplica aquí: https://www.creadorescapital.com/Redes de Juan Pe:- https://www.instagram.com/juanpe.divisual/- https://www.youtube.com/channel/UCh0yfHiqN7Rt0jKlOKYjYXA- https://www.youtube.com/@JuanPeNavarroTrasLaPantallaCreadores- Facebook: https://www.facebook.com/creadorespodcast- Instagram: https://www.instagram.com/creadorespodcast- Instagram: https://www.instagram.com/chelozegarra- TikTok: https://www.tiktok.com/@marcelozegarrac- Twitter: https://twitter.com/chelozegarrac- Email: https://www.creadores.co/contacto#InteligenciaArtificial #AgentesDeIA #FuturoDelTrabajo #NegociosDigitales #CreadoresPodcast

Rana Gujral is the former CEO of Behavioral Signals and the author of the upcoming book The AI Instinct. During his time leading Behavioral Signals, Rana has led a company built on a contrarian bet: that the words in a conversation are the least interesting part of it, and that the real signal, intent, trust, stress, deception, lives in how something is said rather than what is said. His team has mapped roughly 75 to 100 behavioral dimensions in the human voice, work that now powers everything from deepfake detection for government agencies to a matching system that pairs call center customers with the agents they are most likely to have a natural, flowing conversation with. Liam and Rana dig into the unconscious vocal tells we all give off, why pitch compression, not raised volume, is the real signature of suppressed stress, and how studying voice for eight years changed the way Rana himself talks and listens. They also get into the ethics of emotion AI, including why the EU has banned it from workplaces, and the central idea behind Rana's book: that large language models are missing an entire axis called experience. Rana introduces his concept of Artificial General Experience, or AGE, and makes the case that the real fork in the road for AI isn't intelligence versus replacement, it's whether these systems make us more ourselves or less. Key Topics Covered Why Behavioral Signals bet on voice as a behavioral signal instead of a language signal The 75 to 100 dimensions of emotion, intent, and cognitive state hidden in a voice How deepfake detection works when a synthetic voice is good enough to fool a mother Surprising commercial uses of voice AI in marketing, call centers, and fraud detection The science of compatibility: why some conversations click and others feel like effort What 8 years of studying voice changed about how Rana communicates The unconscious vocal tells that give away hesitation and suppressed stress Why the EU banned emotion AI in workplaces, and where Rana draws his own ethical line The idea behind Rana's book, The AI Instinct, and why AGI is the wrong question to ask Artificial General Experience (AGE): the missing piece between intelligence and judgment Where augmentation ends and replacement begins, from GPS to Neuralink The geopolitical inequality of who gets access to the most capable AI tools Episode Timestamps 00:00 - Introduction 00:09 - The contrarian bet: voice as a behavioral signal, not a language signal 02:51 - Mapping 75 to 100 dimensions of emotion, intent, and cognitive state 06:07 - Commercial uses beyond law enforcement: marketing, call centers, fraud detection 10:35 - The science of compatibility and conversational entrainment 13:35 - Beyond voice: body language, physiology, and why voice is the primary channel 16:42 - How 8 years of studying voice changed Rana's own communication 19:32 - The unconscious vocal tells everyone gives off 23:40 - Pitch compression: the real signature of suppressed stress 23:53 - The ethics of emotion AI: EU AI Act, modulation vs manipulation 27:08 - Why Rana wrote The AI Instinct 30:04 - Artificial General Experience (AGE) and what LLMs are missing 33:59 - How lived experience dynamically updates memory and meaning 38:04 - Why the goal isn't to build machines that are more human 40:50 - Augmentation vs replacement: where the line gets drawn 43:55 - Purpose, meaning, and the risk of frictionless cognition 51:23 - What we should be teaching the next generation 56:01 - The geopolitical inequality of AI access 59:11 - What Rana hopes readers take from The AI Instinct 1:02:32 - Where to find Rana and the book Rana's website: https://ranagujral.com/ Rana's LinkedIn: https://www.linkedin.com/in/ranagujral/ Partner Links Upgrade your AI toolkit: https://www.theaireport.ai/ai-executive-pass Subscribe to our free newsletter: https://newsletter.theaireport.ai/subscribe Join the community: https://community.theaireport.ai/checkout/the-ai-report-welcome-gift?coupon_code=WRTH Learn more about your ad choices. Visit megaphone.fm/adchoices

This Week In Startups is made possible by: Northwest Registered Agent- NorthwestRegisteredAgent.com/twist Odoo - Odoo.com/twist MongoDB - MongoDB.com/ai   Today's show: The FCC's decision to ban Chinese humanoid robots over security concerns is a boon to American startups, which now face a narrower competitive market. But where should we draw the line on security over competition? Menlo Ventures' Deedy Das, Weisburd Pierce's David Weisburd, Plexo Capital's Lo Toney, and LAUNCH's Jason Calcanis broke down how they differentiate between legitimate security concerns and purported regulatory capture. Today's venture capital roundtable also dug into OpenRouter's possible sale to Stripe, changing tokenomics, the Indian market for startups, and even DoorDash's drone-delivery business taking flight. Guest Links: Deedy Das https://x.com/deedydas Menloe Ventures https://menlovc.com/ Lo Toney https://x.com/lo_toney Plexo Capital https://www.plexocap.com/ David Weisburd https://x.com/DWeisburd Weisburd Pierce https://www.weisburdpierce.com/ Jason Calcanis https://x.com/Jason LAUNCH https://launch.co/ Show Links: The FCC's decision regarding Chinese robots https://www.fcc.gov/document/fcc-adds-foreign-produced-power-inverters-and-robots-covered-list-0 Pacing the Frontier letter https://www.pacingthefrontier.com/ Cursor Start https://cursor.com/blog/cursor-start-india Stripe may buy OpenRouter https://www.axios.com/2026/07/24/stripe-openrouter-merger-ai-currency OpenRouter financials https://www.theinformation.com/articles/openrouter-financials-suggest-steep-price-possible-acquirer-stripe?rc=g3wfdp Kimi K3 license https://huggingface.co/moonshotai/Kimi-K3/blob/main/LICENSE Pangram https://www.pangram.com/ Tau Robotics https://www.tau-robotics.com/ Zipline https://www.zipline.com/ Manna https://www.manna.aero/ Kindred Ventures https://kindredventures.com/ Autolane https://goautolane.com/ Salmon Labs https://salmonrun.ai/   Timestamps: 0:00 The FCC bans Chinese humanoid robots 2:18 Waymo, Uber, Robotaxi, and the global BYD threat 10:30 MongoDB - AI-assisted and agentic coding is helping you build faster than ever. Start building at https://MongoDB.com/ai 15:45 The "Pacing The Frontier" letter (1,100+ AI staffers warn on RSI) 19:51 Odoo - The all-in-one business platform. Get started for free at https://Odoo.com/twist 21:07 Regulatory capture vs. genuine concern 30:09 Northwest Registered Agent - Get more when you start your business with Northwest. In 10 clicks and 10 minutes, you can form your company and walk away with a real business identity — Learn more at https://northwestregisteredagent.com/twist 37:57 Have we reached AGI? 38:35 Stripe eyes OpenRouter at $10B 39:19 Does OpenRouter have a moat? 47:13 Kimi K3's license and neocloud margins 49:34 Claude Tag and the future of AI-mediated workplaces 53:26 Cursor Start, ChatGPT Go, and the India market 1:02:03 Have we solved AI detection? 1:11:20 Tau Robotics $30/hour robotic housecleaning 1:12:49 Job loss, and the social safety net 1:18:25 DoorDash Air takes on Zipline, Manna 1:21:00 Portfolio shout-outs Subscribe to the TWiST500 newsletter: https://ticker.thisweekinstartups.com Check out the TWIST500: https://www.twist500.com Subscribe to This Week in Startups on Apple: https://rb.gy/v19fcp   Follow Lon: X: https://x.com/lons   Follow Alex: X: https://x.com/alex LinkedIn: ⁠https://www.linkedin.com/in/alexwilhelm   Follow Jason: X: https://twitter.com/Jason LinkedIn: https://www.linkedin.com/in/jasoncalacanis   Great TWIST interviews: Will Guidara, Eoghan McCabe, Steve Huffman, Brian Chesky, Bob Moesta, Aaron Levie, Sophia Amoruso, Reid Hoffman, Frank Slootman, Billy McFarland   Check out Jason's suite of newsletters: https://substack.com/@calacanis   Follow TWiST: Twitter: https://twitter.com/TWiStartups YouTube: https://www.youtube.com/thisweekin Instagram: https://www.instagram.com/thisweekinstartups TikTok: https://www.tiktok.com/@thisweekinstartups Substack: https://twistartups.substack.com

People are calling the new ChatGPT Voice their “AGI moment.”

Jerry Tworek led reasoning at OpenAI, convinced that scaling reinforcement learning was the path to AGI. Rohan Anil co-led Gemini pre-training and built the Shampoo optimizer. Now they've teamed up at Core Automation on a contrarian premise: the transformer has carried us as far as it can, and the bottleneck to smarter systems is no longer scale — it's the architecture itself. The missing capability is continual learning, models that adapt at test time, which transformers can't do. In-context learning taps out fast (Codex needs compacting after ~20 minutes) and fine-tuning invites catastrophic forgetting. Rohan argues pre-training and RL should be optimized end-to-end, and that transformers spend computation inefficiently. They lay out why the largest labs won't chase alternatives while locked in the coding-agent race, and why building the world's most automated lab starts with automating kernel generation—the one place frontier models still lose to a high-taste human. Hosted by Sonya Huang and Pat Grady, Sequoia Capital

In this episode of The Delphi Podcast, Tommy sits down with Travis Good, co-founder of Ambient, to discuss why open-source AI may ultimately beat the closed labs and what that shift means for developers, businesses, and the broader AI economy.Travis explains how Ambient is building a decentralized marketplace for AI inference, matching demand with underutilized GPU capacity while verifying that users receive the exact model quality they paid for. They also explore the risks of building on closed AI infrastructure, China's growing advantage in open-source models, and why OpenAI and Anthropic may be creating long-term distrust among their own customers.The conversation also goes deeper into what would happen if OpenAI achieved AGI first, why prompt injection remains one of AI's most important unsolved problems, and how the current AI infrastructure spending boom could eventually trigger a broader funding shock or AI winter.Timestamps00:00 Intro03:20 What Ambient Is and How It Works18:40 Verified AI Inference40:20 China and the Open-Source AI Race49:00 Why Closed AI Is Making a Mistake1:03:00 What Happens If OpenAI Reaches AGI?1:17:10 The AI Bubble and a Possible AI WinterTommy: https://x.com/Shaughnessy119Travis: https://x.com/IridiumEagle

There are roughly 100x more people who use code than who can write code. As code that “just works” becomes easier to generate, this group may be the biggest prize of all — if you can get the agentic interface right.A key trend we have been tracking over at AINews is the absolute explosion in Codex usage this year, with MAU now up >10x from Jan 2026. Less than two weeks after their July 9th launch, OpenAI said ChatGPT Work and Codex had reached 10M users combined (as we cover in the pod, Codex now powers ChatGPT Work, so all ChatGPT Work users are now users of the Codex harness, even if they aren't traditional engineers) — showing the early innings of what happens when you graduate from coding agents to knowledge work agents:We've been calling out how coding agents are “breaking containment” to do everything else this year to power every other part of knowledge work - and it started with the org chart, with a major reorg last month that amounted to two of Codex's most prominent leaders, Greg and Tibo, taking responsibility over product and ChatGPT specifically, completing a “Superapp” consolidation cycle first discussed in March.With these updates Codex is no longer just a coding tool. In June, OpenAI said knowledge workers already accounting for roughly 20% of Codex's user base and growing more than 3x as quickly as developers. A product dedicated for knowledge workers was being pulled out of the Codex team.However, knowledge work has a different set of problems and environments than coding. For decades, knowledge work has been scattered across different primitives like documents for writing, spreadsheets for analysis, slide decks for communication, and specialized applications for everything else. ChatGPT Work now enables users to work across every primitive with agents. Instead of opening an application and manually operating its features, the user can describe an outcome and collaborates with an agent that can assemble the tools, context, and artifact needed to reach it.From building no-code products at Airtable to leading Productivity Engineering at OpenAI, Akshay Nathan has spent much of his career trying to make the power of software accessible to people who do not write code. In this episode, Akshay joins swyx and Vibhu to unpack the launch of ChatGPT Work, why Codex unexpectedly took off among non-developers inside OpenAI, and the company's broader plan to bring useful agents from software engineers to knowledge workers and eventually everyone.We go deep on the shared agent harness behind Codex and ChatGPT Work, why OpenAI brought the experiences together without making them identical, and how persistent computers, artifacts, Sites, plugins, memory, and sub-agents are changing what people can delegate to AI. Akshay explains why some teams are replacing decks and spreadsheets with interactive websites, how agents can gather context across code, Slack, documents, and local files, and what OpenAI learned from personal-agent products like OpenClaw.Side note: also don't miss Abhihek's sandbox track keynote at AIE, which now powers a lot of the sandboxing for ChatGPT Work… and yes was also broken by an unreleased OpenAI model in the recent HuggingFace incident.Akshay also reflects on how AI is transforming product development itself: why more people will become generalists with a specialty, why ideas and taste become the bottlenecks when almost anyone can build, why LLMs still struggle to generate genuinely grounded new ideas, and why teams must distinguish increased motion from actual progress.We discuss:* Why Codex unexpectedly took off among non-developers inside OpenAI* Why employees felt like using Codex gave them a new superpower* The product insight that led OpenAI to build ChatGPT Work* Why Codex and ChatGPT Work share the same underlying agent harness* How their UX, Git visibility, artifacts, and sandboxing defaults differ* Why OpenAI merged its agent experiences instead of building separate products* How AI is blurring the boundaries between engineering, design, strategy, and operations* Why OpenAI wants the default model configuration to work for most users* When power users should use deeper reasoning, Ultra, or multi-agent modes* Artifacts, agentic spreadsheets, and creating high-fidelity work products* Why interactive Sites may replace decks and spreadsheets* The challenge of designing a simple interface for an agent that can build almost anything* Why users should retry tasks that models could not handle three or six months ago* How AI can gather context for performance reviews without replacing human judgment* The OpenAI automation that turns internal Slack and document activity into memes* What reaching ten million ChatGPT Work and Codex users means for the product* How OpenClaw inspired persistent environments, scheduled tasks, and personal agents* Using ChatGPT for financial planning, budgeting, workouts, meals, and household management* The design tradeoffs behind sub-agents and how much of their work users should see* ChatGPT memory, Chronicle, and long-term context* Why AI may make more people generalists with deep specialties* Why ideas and taste become more important when almost anyone can build* Why LLMs still struggle with the instruction “bring me new ideas”* Measuring productivity through quality at-bats instead of commits, tokens, or pull requests* The critical difference between AI-generated motion and meaningful progressAkshay Nathan* LinkedIn: https://www.linkedin.com/in/akshaynathan/* X: https://x.com/akshaynathan_Timestamps00:00:00 Introduction and Bringing the Power of Code to Everyone00:01:33 Joining OpenAI and Preserving a Startup Culture00:02:40 What OpenAI Learned from Enterprise AI Adoption00:05:28 Why OpenAI Built ChatGPT Work00:07:17 Codex vs. ChatGPT Work and the Shared Agent Harness00:12:07 Why OpenAI Merged Its Agent Experiences00:16:24 Models, Reasoning Levels, and Choosing the Right Default00:20:26 Artifacts, Agentic Spreadsheets, and Model–Product Collaboration00:24:22 Why Sites Could Replace Decks and Spreadsheets00:30:08 Designing an Agent That Can Build Almost Anything00:34:28 From Developer Agents to Knowledge Work—and Everyone00:36:07 Power-User Advice and AI-Assisted Performance Reviews00:40:41 OpenAI's Internal AI Memes and the Ten-Million-User Launch00:44:39 OpenClaw, Personal Agents, and ChatGPT as an Operating System00:50:24 Sub-Agents, Ultra Mode, and How Much Control Users Need00:54:39 ChatGPT Memory, Personalization, and Chronicle01:00:19 How AI Is Reshaping Product Development and Tech Roles01:03:15 Ideas, Taste, and Why LLMs Struggle to Generate New Ideas01:04:42 Measuring Productivity, Quality At-Bats, and Motion vs. ProgressTranscriptIntroduction: Akshay Nathan, ChatGPT Work, and the No-Code ArcSwyx [00:00:00]: We're here in the studio with Akshay from OpenAI. Welcome.Akshay Nathan [00:00:07]: Thank you.Swyx [00:00:08]: And with our trusty co-host, Vibhu. So you recently launched ChatGPT Work. You lead Core Product Engineering. It's been a long journey, into all this. I find it very interesting that you started with no code or low code, with Walrus and Airtable. And to some extent, ChatGPT Work is like the super app of super apps of, well, here is the ultimate no code. You just write a prompt.Akshay Nathan [00:00:32]: Yeah. It's funny how things come, full circle. I think for a long time in my career, I started my career working consumer fintech, but then after that, like, there's this hypothesis that, the things that we were able to do with code, like, as engineers, like, if we could bring that to many more people in a more, accessible way, then that would be truly magical. We were working on a startup. It's funny, like, before LLMs, before vision LLMs, on how to do automated testing with AI. It was just kinda jank, back then, but doing what we can, and then worked at Airtable for a while on the same thesis that, like, if we can bring a database or the primitives behind a database to people, that'd be really useful to them. But once LLMs came onto the scene, it became clear that, this was the missing piece, like, the missing technology required to, like, bring the magic of code to everyone without them having to know what's going on underneath the hood. And so, like, I think this launch and a lot of the stuff that we've been up to is, like, the manifestation of that.From Walrus and Airtable to OpenAIVibhu [00:01:33]: How was stuff when you joined? So you joined OpenAI 2023. Now we've got, so much more stuff, so ChatGPT, Codex app, ChatGPT Work. Have things changed?Joining OpenAI and What Hasn't ChangedAkshay Nathan [00:01:44]: I think the more interesting thing is how things haven't changed. Like, one, I joined I remember when I joined, it was, like, five hundred people. One thing I was worried about was, like, I was looking for something, more early stage and, like, was it gonna feel startup enough? And I joined, and I was like, “This feels even more startup-y than I could ever imagine.” And, like, that really hasn't changed even till now. I think the, like, level of, like, bottoms-up ambition and, like, the ability of anyone to, like, do anything or have an idea and ship it is really cool. But on the, like, mission side, I think what was really compelling to me is this mission of, bringing frontier intelligence to everyone. Like, building AGI and then bringing it to everyone. And, I think acknowledging back then that, like, that vision is gonna, not be a linear progression. Like, we're probably gonna, like, try different products and have different things that succeed and don't. But the vision has stayed the same, and the mission has stayed the same, and we're starting to see the pieces, fall together, and that's really cool.Enterprise Lessons: No One-Size-Fits-All AISwyx [00:02:40]: You worked on Enterprise. What A lot of people never touch ChatGPT Enterprise. What is something that you learned from there that you're bringing into your work now?Akshay Nathan [00:02:52]: I think how there's no one-size-fits-all solution in Enterprise. I remember in the early days of ChatGPT Enterprise, like, when we talked to customers and, like, everyone. That was, like, when I think it was a year after ChatGPT was released, and everyone was so excited to bring, AI into their enterprise. And, there were all these teams being stood up. It was, like, the AI deployment team with, like, these enormous budgets. And if you asked anyone, like, what were they excited about? Like, what were they excited about solving? Like, at first, you'd get, like, kinda like the baseline answers of, like, “Yeah, we have all this context and data and all this stuff.” But then if you ask them, like, “What was, like, a discrete use case that, like, they want AI to enable in their workplace?” You get such a different, like, variance, like, explosion of, different types of answers. And it's interesting, like, you using, like, these models and these products, you have this box, and you can say anything to it, which is the magic. But it'on the flip side, it also means that, like, you don't know what to do with it. And in Enterprise, I think a big part of that is, like, meeting the users where they are, like, what use case were they trying to solve, and then teaching them how they can use AI to, like, gain leverage there.Swyx [00:03:56]: Do you meaningfully differentiate that from forward-deployed engineering?Akshay Nathan [00:04:01]: I think there is the go-to-market side of it and then there is the product side of it. I think you need someone on the product side. And I think, like, however good we get at FDE motion, like, I think at the end of the day, if we have a user who's, like, looking at their computer or looking at their phone, like, it's our job in the product to, like, be enabling them and showing them where to go. So we're really excited about that.Vibhu [00:04:24]: Do you think there's been changes, over the past three years of adoption? So there have been, step function changes. You have reasoning models and whatnot. Is there still the same problems of Enterprise has black box, don't know what to do with it, or have things changed?Adoption, Agents, and the Next 10x MarketAkshay Nathan [00:04:39]: We're seeing now that, like, there's this huge uptake, right? Everyone is extremely excited about it. It feels like, many people are, millions, hundreds of millions of people are using ChatGPT. They understand, like, how generally to work with AI. But then, like, every time, like, a new capability gets unlocked, so now, like, we're seeing with agents, like, there is probably a contingent of, like, early adopters still who, truly get it, who are like, “ we you can do anything. You just have to make sure the right context is there, it's connected to the right tools, and that you are supervising it, but, like, anything is possible.” But then there's, like, this, like, 10x or 100x bigger market where, like, they don't yet get that, or they don't yet see that. And so I think that's the next stage here. So to answer your question, like, I think the adoption is there and growing fast, but I think the opportunity is, like, far bigger than that. That's where we wanna play, especially with ChatGPT Work.ChatGPT Work, Codex, and the Super App MergeSwyx [00:05:27]: Yeah. well, let's, let's skip ahead to ChatGPT Work. only, like, a month ago or so, announced. what was the decision process that led into it? there was this, overall merging of the super app. Is that what we're officially calling it? you deprecated the browser as well. Just, summarize your last, like, couple months of working on this thing.Akshay Nathan [00:05:50]: Yeah. It feels like forever now, but it's only been a few months. I think maybe the one, impetus that, like- Is most salient is when we release Codex, or even internally had Codex, like, it was really surprising to us, I think we recently put out some stats on this, that there was this, like, real inflection of, like, adoption among non-developers at OpenAI. And, I, through this product development process, like, would go to, like, these UXR sessions to talk to people internally. And the thing that stuck out to me is, like, one, like, you go talk to, like, strategic finance or marketing or whatever, and they're all using Codex for, their use cases. That part's cool, but the thing that really stuck out to me is how proud people were that they were using Codex. Like, how, likeSwyx [00:06:34]: It's like, “I'm not supposed to be using it, but I am.”Akshay Nathan [00:06:36]: It was that. It was, like, that they were, early to this, like, new thing, but it was also this thing of, like, they felt like they had a superpower, right? And, what we recognized then is that, like, the power of Codex, the power of agents, like, we already had this massive distribution base of people who have, come to know and love ChatGPT. Like, how do we show that to them? Like, how do we bring it to them? Which is, like, a hard product problem, and it's, like, a tricky thing, right? There's many ways you can go about it. And so that's what we called the Merge and the Super App over time, and ultimately launched it in ChatGPT Work, is how do we do that? But it came from that initial realization that, like, the power was not only for developers, like, much earlier than probably even we thought. Like, it could be extended to everyone.Swyx [00:07:17]: How do you see the products differently? So, like, who is it for, right? So Codex started out even CLI, then app. Now there's a merge of ChatGPT Codex and ChatGPT Work, so is it the opening for the average user, for enterprise, for work? How do you position it?Akshay Nathan [00:07:36]: I think we want to get it to position it for if you're doing work-related things, for lack of a better word, right?Who ChatGPT Work Is ForAkshay Nathan [00:07:42]: I think productivity is what, like, the pillar that I support. Like, that's the name of the team. And the reason for that, the reason we call it productivity and not, like, enterprise or, like, work or something like that, is because there's also personal productivity, right? And, like, I think ChatGPT Work is I've seen people do things in their personal lives that you wouldn't classify as, like, work technically, but, like, these agents are, super capable for. Like, one recent example that someone posted about, on our Slack is, like, someone had, like, a missed package, like they didn't receive it, and then they got, like, the picture of it, from Amazon or whoever the courier was, and they, like, asked ChatGPT Work to, like, find out where that package is. And, like, the agent, is extremely tenacious and, like, took the image and, like, looked at a bunch of, like, listings around their neighborhood and figured out exactly the apartment complex in which the package was, like, gave them some information. And so, like, I think there's all these things that, like, you, work-related or productivity-related things, I think that's what we want the product to be. You asked about Codex. I think we think Codex is, a durable brand, but we have a principle that, like, the user we don't want a user to get stuck in a tab or an experience where they don't get the power of the product. And so, like, everything that you can do, in the Codex portion of the product on desktop, you can do in ChatGPT Work and vice versa. But we made some opinionated product decisions on, like, how much of the Git state, if you're in a Git repo, do we wanna expose to the end user? Or how much do we wanna make the experience of seeing the agents thinking, like, diff forward so that you get exposed to the diffs out of the box. And then, like, on the safety side, like, how do we wanna think about, like, sandboxing and making sure that we have the right defaults in one state versus the other? So, there's, like, some opinions that go behind that, but we do want We don't want the user to need to choose which experience they're in.Swyx [00:09:26]: That is a good goal for AGI, right? Like, people don't want, like, to hide to choose what version of AGI they want. They just want the AGI to decide for them. can I get an answer or, like It's not super clear to me. Is the Codex harness and the ChatGPT Work harness the same? Is it just UI affordances, or are there prompt level or even deeper differences?Shared Harness, Different UX: Codex vs. WorkAkshay Nathan [00:09:49]: So the harness is the same. The harness is shared. on In both of the products, we made improvements to the harness to make it good for knowledge work, especially as it relates to plug-ins or computer use or artifacts. You get that power regardless of which experience you're in. On the UX side, there's opinionated takes that we have when you're in Codex mode, what the UX should be how the UX should behave, and some stuff around the sandbox like I mentioned, but the underlying harness and capabilities should be the same.Swyx [00:10:16]: I'm just kinda curious. Maybe we can, -- Is there a query that we can run that would look different in the two modes?Akshay Nathan [00:10:23]: Yeah. I tried to create, like ask it to create, like, a retirement calculator spreadsheet or something, in both modes. And then in Codex mode, you might have to be in a repo for this, but you'll see, like, the diffs of, like, the sheet that it's creating and stuff like that, and the file edits. But in Work you won't be able to see that.Swyx [00:10:42]: I think that's, that's super clear. And then also the other thing I wanted to dive into was your, the productivity team. what else is there? first of all, what are the top-level teams other than productivity? Isn't productivity everything?Productivity Teams and Core ChatAkshay Nathan [00:10:55]: SoSwyx [00:10:55]: Science?Akshay Nathan [00:10:55]: We have a team focused on ChatGPT. Like, the core chat experience, for consumer, which is like, not, I think all productivity. Like, there'People are using ChatGPT every day for search to, figure out how to write messages to loved ones, to think about, how to, like, learn a new topic, et cetera. And so there's so much more inside to create images. And there's so much more in chat that, the hundreds of millions of users are using that warrants, like, a very dedicated effort. And there's teams focused on enterprise and infrastructure and API and stuff like that, so.Swyx [00:11:33]: I will bring it up.Retirement Calculator Demo and Git-First UXSwyx [00:11:34]: Yeah. So I have them both running. This is ChatGPT Work. There's a Codex version here. I picked “Five Little Ducks” song, so this will take a while.Akshay Nathan [00:11:43]: Huh.Swyx [00:11:43]: I think we'll just keep it in the background and, as they finish, we'll look into some of the differences.Akshay Nathan [00:11:48]: Yeah. But immediately, I think if you flip back to the Codex version you'll see that,Swyx [00:11:53]: That it assumesAkshay Nathan [00:11:54]: Like theSwyx [00:11:54]: It assumes Git. Yeah. Yeah.Akshay Nathan [00:11:56]: The, like, dynamic island assumes that you're in a Git repo. And you might miss some stuff because some of it is, like, in the actual chain of thought with those changes and how we display that, but yeah.Swyx [00:12:07]: Is there an unintuitive like, is there a thing that you wanted to ship and then you got feedback, and you were like, “No, let's not do it?” Like, what's the thinking behind that?Why Merge the ExperiencesAkshay Nathan [00:12:14]: In, ChatGPT Work?Akshay Nathan [00:12:17]: I think one direction we could have gone with this is, like, keeping the experiences, like, completely separate. So it's like, whySwyx [00:12:22]: Different apps.Akshay Nathan [00:12:23]: Exactly, like different apps or even in the same app, like different, completely different experiences. Like, why merge it all? Like, what is. Codex, people love. Like, why bring these products together? And I think the intuition here is that, like, all of our jobs are, like, changing dramatically with AI. Like, for, like, every few months, like, I feel like I wake up, and I'm, like, doing a completely different thing than I was doing a few months ago. And my hypothesis here is that, or I should say our hypothesis is that, like, part of what we're, we're building, this technology is giving people leverage. Like, the things, maybe it's the more mundane parts of your job or parts that, like, if you were able to automate, you'd be able to share more ideas faster or whatever, like, you're able to do now. And because of that, like, that might blur the lines between someone who's, like, only writing code or creating strategy docs or, planning events or, helping with marketing or doing podcasts or whatever, right? And so, like, these things are gonna get blurred over time. And so, like, trying to draw a hard boundary based on, like, the who you are is gonna be, is gonna be tough. And, like, we should enable users to choose, but we shouldn't box them in. And so a lot of the work that went in here, like, keeping the primitives the same, like for example, plugins are, like, unified across, this product and ChatGPT and the cloud, was because of that. It's this thesis that, like, eventually things are gonna come together and we don't wanna be Like, we wanna be prescriptive about when to be in either experience, but we don't want to box anyone in.Swyx [00:13:45]: I wonder if there's users who are very tuned to the old ChatGPT harness that is effectively now replaced by the Codex harness. I can't imagine what that was, but maybe they're more the more conversational side. Can you compare and contrast the two harnesses? ‘Cause only you've seen it.Akshay Nathan [00:14:02]: Yeah. I think ChatGPT, the existing harness, like, still exists today. Like, it exists in this app,Harness Engineering: ChatGPT vs. CodexSwyx [00:14:08]: The classic, right?Akshay Nathan [00:14:09]: TheVibhu [00:14:09]: You just start a new chat, and you don't go under Work, right?Akshay Nathan [00:14:13]: Yeah. If you startVibhu [00:14:13]: SoAkshay Nathan [00:14:14]: A new chat and go to chat, then you're, you're talking to ChatGPT with the instant model.Vibhu [00:14:16]: Oh, we can technically do another. But on instant.Swyx [00:14:21]: Yeah. So this one's not gonna code or it's gonna be in line. It's on a in line in a sandbox.Akshay Nathan [00:14:26]: It'llVibhu [00:14:27]: Oh, that's coolAkshay Nathan [00:14:27]: We try to push you to go to Work if you're creating a spreadsheet. Yeah, but this isSwyx [00:14:30]: And this is a router decision? Sorry. Is it a router decision?Akshay Nathan [00:14:34]: This is the decision that, the model is making, and then, like it sees that you're able to. or you're trying to do something that would be better served in Work mode. But I think your question was like, what are the advantages of, like, the chat, like ChatGPT chat harness?Swyx [00:14:48]: It's more broadly, like, I wanna, do an oral history of harness engineering. Right? the ChatGPT harness lasted us from, let's call it the ‘01 era, until now, and now it's being replaced by the Codex harness effectively. And they're, they're overlapping somewhat, but I'm curious what changed if there is.Akshay Nathan [00:15:10]: My perspective on this is, like, there's, there's, there's there's like a constant process of, like, divergence, convergence, divergence, convergence. And in chat, like, many of the use cases I was talking about before, like, search or learning, I think we're, we're really optimizing for latency and optimizing for personality and, like, different things that, over time, like the product The reason people love ChatGPT is because we've been optimizing for those things and working on them for so long. Codex, what we learned was that, like, if you give the agent access to this infinitely flexible environment as a computer, it can do really powerful things. And so when we think about, like, okay, well, for knowledge work, like, what is which mode should we choose? It was like it felt more natural to us to bring that to this, like, computer environment and, maybe abstract some of the details of this computer away from users who might not be used to that, but, like, give them that same power. But ultimately, I think that we want the power in all places, right? We wanna meet people where they are. So I'm sure there'll be work down the road in order to get things to be, equivalently capable in all scenarios. But it's just a question of, like, what we've been focusing on the product on historically and what we're focusing on now.Models, Defaults, and the Reasoning SliderVibhu [00:16:24]: I think alongside that, outside of just harness and when to use Codex, ChatGPT, or Work, there's also the new models you've released, right? any guidance there? So people love to min-max what to use, like only use Terra on high reasoning versus, for this, you wanna use Sol here, ignore all theseAkshay Nathan [00:16:44]: There's 32 options.Vibhu [00:16:46]: But, that being said, for people that are expanding, so, productivity trying stuff for work that don't have the breakdown of what all this is what's, what's the advice, right?Akshay Nathan [00:16:59]: Well, I think before the advice, like the first thing is, like, none of this would be possible without these models. Like, the, I think you asked earlier, like, what was, like, the inspiration for work and, like, early on, like I mentioned, like, what we were seeing with Codex, but that was also because the models were getting infinitely more capable. That's happening again. I think it's like another step function jump now. And to answer the question on advice, like we want this default to be the best possible. Like, we wanna be opinionated about the default, and so we've we've chosen a default that we think is gonna be the best for everyone. And, we have for power users options under the hood. We could One could argue that there might be too many right now, and we're, working on simplifying it. But you can extend, the reasoning level, and you can change between the different model classes if you need to, but the default should be the best for most use cases. So my advice to most people would be to stick to that. And then, if you reach a situation in which you think that you could, you wanna try, a different configuration, if you're not seeing either the efficiency on the cost side or the quality on the intelligence side, then you can change the defaults and see if you can get something better. But we think that the default should be good enough.Swyx [00:18:09]: I have, I'm just gonna run something by you since you have way more experience than me. I've recently been doing Sol Lite but with goal, with the idea that the goal augments the reasoning effort, but with more terminations and turns.Swyx [00:18:24]: Is that a good way to think about it as opposed to Sol Ultra or Sol, Extra High?Akshay Nathan [00:18:29]: Yeah. It's hard to say becauseSwyx [00:18:31]: Yeah. It's like an interaction effect.Akshay Nathan [00:18:33]: exactly. It's like there's a preference on, for you as an individual, like how do you like to collaborate with the models? Like how many of those like terminations, as you call them, do you want where, you can steer or make sure that it's doing the right thing?Akshay Nathan [00:18:46]: I think generally people should try whatever works for them. I think that like using Ultra or the like multi-agent setups are best for like when you have like tasks that are either incredibly complicated, like open explorations or very paralyzable. I think even for tasks using goal, I think is best for tasks that you'll be able to make consistent progress in a way that's verifiable over time. But I think for most tasks, they don't fall into either of those buckets. And so like at least when they're starting, and so that's why I think the best first step is like trying it with the default configuration and then seeing like where you wanna go from there.Swyx [00:19:29]: Right. You guys worked on a slider, which is super helpful for reducing the amount of panic.Vibhu [00:19:36]: It's nice on mobile at least. There's a nice slider there.Swyx [00:19:38]: It's nicer.Vibhu [00:19:39]: I haven't tried it.Swyx [00:19:40]: So you have the advanced view there, but if you click advanced view. Yeah.Vibhu [00:19:44]: Ooh, it's just a nice slider. Yeah.Swyx [00:19:46]: Very pretty, very colorful.Akshay Nathan [00:19:48]: Yeah. The idea was here was like reduce it to like one dimension even though there's multiple dimensions, right? Try to project it onto a single dimension for the user. Like, something from that represents like, speed and efficiency on one side and then like quality and thoroughness on the other side.Artifacts, Spreadsheets, and the Work LaunchSwyx [00:20:04]: I am just puzzled that it uses Sol so much, like the lowerVibhu [00:20:07]: NoSwyx [00:20:07]: Grounds I would've usedVibhu [00:20:08]: I think the slider, if I'm not mistaken, isSwyx [00:20:09]: Terra.Vibhu [00:20:10]: Oh, it is.Swyx [00:20:11]: Yeah. See? So they preset Terra to only be the light one. But like I think a lot of people would more people should use Terra. One, because Sol keeps running out of capacity.Vibhu [00:20:22]: I'm the reason. Here's ten minutes of ourSwyx [00:20:24]: There you goVibhu [00:20:25]: Retirement calculator.Swyx [00:20:26]: Oh, that's the Excel thing working for you.Vibhu [00:20:28]: This is,Swyx [00:20:28]: Oh my God. Look at thatVibhu [00:20:28]: This is work, and then Codex is still cooking, so we'll get back into it. I think it'll be interesting to see the thought process, the reasoning, and also, this is eight minutes on work. Codex is still cooking.Swyx [00:20:41]: Yeah. And by the way, so I've, do Gabriel Chua? He's part of the OpenAI Singapore team. He showed me this, and I was like pretty shocked that this looks like Excel. It edits Excel files. You never paid an Excel license, right? Like, but somehow this is like workable and it's agentic Excel.Akshay Nathan [00:21:01]: Yeah. one of the big like pushes that we made for this launch was like artifacts, right?Akshay Nathan [00:21:05]: Like both on the model side, like I think if you compare this with GPT-5.5 and GPT-5.4 before that, you'll see that there's been pretty dramatic improvements in the quality of these artifacts and then also on the product side.Vibhu [00:21:16]: The UX side is also crazy, like hosted sites and whatnot. No longer needing to host your own little webpage, like itSwyx [00:21:23]: Oh, I have a story about that. I can do, a separate thing. I'll need to take the visuals here, but we-we'll, we'll cut to that later. Was there co-training, because you were moving making this big move and you launched GPT-5.6 on the same day as ChatGPT Work? Was there influence between the model training teams and the harness teams, or did they did the launch dates just happen to line up the same day?Akshay Nathan [00:21:46]: I think the we collaborate heavily with the research teams, and I think that's like one of the most magical parts of the job, like the most fun parts of the job. But yeah, just using artifacts as an example. Like, a lot of what you're seeing, like underneath the hood, there's a lot of work that went into making sure that like, we had the right infra to be able to train the models to get better at this. And then on the product side, like had the right experience for users to be able to collaborate with the model on an artifact like this. In fact, like this whole viewer, like the intuition here is that like, it's not necessarily that you wouldn't need an Excel license. This is stage one, right? Like, this is probably not what you meant when you're like making a retirement calculator.Vibhu [00:22:24]: Yeah, you can iterate very easily. Yeah.Akshay Nathan [00:22:24]: You wanna iterate and like when you're seeing it, and if this thing is high fidelity to like what you would see in or what your coworkers would see if you were to send this to Sean, like that I think makes it so easier and makes you trust the product in terms of iteration.Vibhu [00:22:39]: When you say coworkers would see, do you see a multiplayer, multi-team collaboration with artifacts? Any things you guys think about that?Multiplayer Artifacts and CollaborationSwyx [00:22:46]: You can already share it, right?Akshay Nathan [00:22:48]: Yeah. It's inter It's something that, we're actively thinking about. one thing that, we've noticed internally without talking too much about the roadmap is that like there's many times when someone will ping me about something, and I will ask ChatGPT Work the question, and then I'll ping them back the answer.Akshay Nathan [00:23:04]: And then I'll be thinking likeVibhu [00:23:04]: Like the simplest would be, the three of us are just all on one hosted.Akshay Nathan [00:23:07]: Exactly. And I'll think about like was I required in this loop or and then maybe it was, rephrase like what they were asking or pulled from certain context or whatever. But like, when I gave them back the answer, that process was also lossy, right? Like I gave them just like my interpretation of what ChatGPT Work cooked up. But like underneath the hood, there's so much context like in the rollout and stuff that could be interesting.Vibhu [00:23:28]: Yeah, it'sSwyx [00:23:28]: So like the answer was preemptively respond to every inbound request?Akshay Nathan [00:23:33]: No, it was just like literally like this is what I do sometimes as my job.Swyx [00:23:36]: I know you copy-paste and then you're just a message forwarding serviceAkshay Nathan [00:23:39]: Yeah. Yeah, exactlySwyx [00:23:39]: From AI to AI.Vibhu [00:23:40]: But I think it's interesting, right? It helps people understand the capability of what you can ask and delegate that oftentimes people don't realize until they try or someone shows you, and then you're like, “Oh, okay. Okay, I see.”Swyx [00:23:52]: I think it's als there's also like a, light security issue, where like you're the permissions layer. Like yes, I could query everything that you query, and I could get an automated response, but maybe I'm not supposed to see it. And that there's no way I would know because I'm not supposed to know what I don't know.Akshay Nathan [00:24:07]: Especially as like, with ChatGPT Work, we're, we're asking you to connect your plug-ins and, it's pulling from your local files and stuff like that. Like the amount of context that the agent has access to is like- Deeply personal and like that's something I think we need to preserve, so that'll be definitely a challenge.Swyx [00:24:22]: There's Excel, there's PowerPoint, there's Docs, the, grand trio of work. What other formats of work do you think about? like you worked on Airtable. Is there a future where there's like OpenAI Airtable? Like what does that look like if you ever ended up doing it?Akshay Nathan [00:24:41]: It's a really good question. I think,Formats of Work: Sites as Knowledge ArtifactsAkshay Nathan [00:24:43]: one that you didn't bring up was Sites, and I think that wasSwyx [00:24:46]: SitesAkshay Nathan [00:24:46]: A core part of this launch. There's one side of Sites that I think people commonly talk about, especially on Twitter and stuff or X, of like, this like prototyping tool. And like we saw that happen with this launch even. The model slider that you guys were referencing earlier, like that was developed almost fully in a Site. Like, the collaboration between design and engineering and product on that was like on a site where we play with, the affordance and figure out how it feels and all of that. But the other aspect that I think is a little bit less talked about is like Sites as like an artifact for knowledge work. I was talking to someone the other day who's on like our corporate finance team, and like we were mentioning how like now when they have these reports that they're, they're working on as a team month to month, historically those things were in slide decks and in spreadsheets, and now they're just in Sites. And like Sites is the mechanism that they collaborate across the team. And the reason is ‘cause it's like, it's like somewhat higher bandwidth. Like, at these tools like PowerPoint and Excel are like infinitely flexible, but at some point you reach the boundary of like either as a human you may not know how to use some feature or something, or the product itself doesn't support it. But with a site you can do anything. You ask for anything and you can get that. once people see that magic, I think it's been really valuable.Swyx [00:26:02]: Yeah, let me show you my case study. this involves all the hot topics including ChatGPT Work, but also GPT-5.6 token billionaires and token maxing and Sites and auto research. I'm a fan of this game called Strata. It's, it's like a little board game that youSites, Auto Research, and Research DashboardsSwyx [00:26:17]: That you play with, physical blocks, that come on top of it like that. So over the weekend I took like thirty photos and just threw into ChatGPT. one point seven billion tokens later, out comes this site with a fully playable thingAkshay Nathan [00:26:32]: WowSwyx [00:26:32]: With 3D, block placement and everything. Because it requires physical blocks and I needed friends to train on it so they can get better, so I can play against them. But also, I could also, do things like train an AI on it and that's, thatAkshay Nathan [00:26:45]: That's your auto researchSwyx [00:26:46]: That gets into auto research. So, you want to train your own AIs, and then make sure they self-play against, each other. I need to set both AIs. So this is AI versus AI, and they're, they're gonna self-play. the AIs start out bad and then you want to define a loss function and get good. I wasn't gonna supervise all this. I was at, I was down in San Mateo, attending a conference. What I ended up doing was, auto researching and on this and creating benchmarks and that there was just way too many parameters for me to read. So I started asking it for a site, and it's created this lab, panel. Where is there a, is there a shortcut for a site that is created?Akshay Nathan [00:27:28]: You should be able to go in the sidebar to Sites, top of the sidebar. The left sidebar.Swyx [00:27:33]: This one? Oh, left?Akshay Nathan [00:27:35]: Yeah. Just scroll all the way to the top.Swyx [00:27:36]: Oh. Oh, it says Sites. Oh, there you go. Yeah.Akshay Nathan [00:27:39]: Ooh.Swyx [00:27:40]: So it create, it creates the sites. I don't, I don't think this is, it is exactly what I wanted, but let me show you what it popped up, right? Like I think as a research artifact, it is very important to communicate, exactly, what is being done. Outputs this thing which I eventually started publishing. So I moved it off of Sites because I wanted more, database and infrastructure than Sites afforded me. But this is like a research output that you can start to mess with and like try to think about like what hyperparameters are you tuning for training AIs. And like I was trying to make like scaling laws and everything and doing all sorts of like game optimization stuff. And the fact that you can just throw this up as a research artifact, like I no longer need to read ChatGPT output. I read Site output. But then there's also a huge sprawl. Like look at how long this thing is. There's so many numbers. It is pretty overwhelming, so then I have to start pruning it from there. But, it's an interesting transition from Markdown effectively that you're putting out to, you're putting out a whole functional site.Akshay Nathan [00:28:41]: I think Markdown just isn't that optimal for people to read, right? Might as well just write HTML website and I don't know. I think you can do a lot with customizing this, right? You have your skills that explain what you want. Like I noticed they're quite verbose. I don't need a lot of this information.Swyx [00:28:57]: It's very verbose.Akshay Nathan [00:28:58]: So and then the nice thing of having a site side by side is, you just iterate on what you want and what you don't, right?Swyx [00:29:05]: Yeah. I don't know if, any that triggers any stories for you of how it's run internally. Am I doing this right?Akshay Nathan [00:29:11]: Yeah. I think that this is like a workflow that we're seeing like all different types of teams use, where like the canonical artifact that was previously a deck or something is now becoming a site. And like with a site you, because it's just HTML, you can like. It's infinitely flexible. And so, if you want to give more prominence to a certain thing that like in a slide deck would, feel like it was buried, like you can do that. You can have it be like the hero image, right? And so I think that like, people are starting to see that. There's more work to be done to make these things like much more easier, easy to collaborate on. You mentioned that they're very, they're long and verbose, could be broken up. I'm sure that there's still something to do there.Swyx [00:29:53]: They're super long. Yeah.Akshay Nathan [00:29:54]: Yeah. But I think we're starting to see that like there is this aspect of this is a really interesting, format, for people to use, that's like much more flexible than what they ever had before.Swyx [00:30:07]: I think your job also comes becomes meta. You're not designing the products. You're designing a product to make products, and I'm curious how you manage that.Designing a Product That Makes ProductsAkshay Nathan [00:30:18]: I think one thing that we've been Like when we look at the UX, like that we've been thinking a lot about is how can we balance like simplicity with capability? Like if we're designing a product, like you said, that like is made to make up build other things, right? You can build so many different things. But we can't put that all in front of you because you'll get overwhelmed.Vibhu [00:30:41]: Yes.Akshay Nathan [00:30:41]: And so we had similar problem or similar challenges even Chat-with ChatGPT, but especially now, like when there's so much that can be done, I think the balance that we're constantly trying to strike is like, how can we give the user enough of a UI surface where, they can be expressive, they can tell the agent what they need, they can verify that it's using the right tools, it's pulling from the right sources, et cetera, but then it gets out of the way. And then how can we build the right system such that we can show them instead of telling them what can be done? Because so much of this is gonna be like, how do they discover the next use case and the next one after that if they really want to be super powered by the AI.Games, Private Evals, and Show-Don'TellVibhu [00:31:19]: Yeah. It's interesting. I feel like everyone also just has a different way to do it, right? I made a similar version of this same game. I didn't take any pictures of board or rule game. I threw in at goal eighteen minutes, fifty-three seconds later, a lot of tokens later, I've got a similar version. not with all the auto research and whatnot, butAkshay Nathan [00:31:39]: You gotta do all the latest trends.Vibhu [00:31:40]: And yeah, I did it with, did it with Codex, not Work, but it's interesting, right?Akshay Nathan [00:31:45]: Yeah. And this is GPT Image generating the pro avatars. Very good for game design. LikeVibhu [00:31:51]: AndAkshay Nathan [00:31:52]: A lot of game designers were like really into GPT Image for assets.Vibhu [00:31:54]: I will say like the broader takeaway probably is the reason that we do this is more so just to test the tools, right? Like, this was also a test for GPT-5.6 came out. I had done the game on GPT-5.5, right? The ability for me to no longer need it to. I had to feed it the rules. It's, it's a pretty niche game. It couldn't find how to do this on its own.Akshay Nathan [00:32:15]: Oh, yeah.Vibhu [00:32:15]: GPT-5.6Akshay Nathan [00:32:16]: It is out-of-distribution, which is why I was also very keen on testing the GPT-5.6 capability.Vibhu [00:32:21]: But, this is just as work comes out, as new things come out, these are just our side ways to test things, right?Akshay Nathan [00:32:27]: Yeah. It's some private eval. That is not this private.Vibhu [00:32:31]: But also valuable because now you can send this to your friends and I learned about this game through seeing this.Akshay Nathan [00:32:36]: It's a hard game. He's very good.Vibhu [00:32:39]: It's good to when no one is competing with you. But yes, it's a classic RL problem of like self-play, bootstrapping your game AI. yeah, you see how easily work becomes personal and personal becomes work because the thing I do for personal, it directly informs people I work with because I showed it to them. They were like, “Oh, you can do that with GPT?” Which like I imagine is the growth strategy.Akshay Nathan [00:33:02]: Yeah. The show not tell is a big piece that, I think we've we're not still not fully cracked of like, showing people all the things that they can do with the product versus like trying to teach that to them through like, articles or onboarding or whatever.Akshay Nathan [00:33:18]: So meeting them in the moment.Vibhu [00:33:19]: It's a career risk for me, because I used to be in developer relations, right? Where your job is to show, and then you're like, “What do you mean? You don't, you don't need.” your job is to tell. And then. But the product people are like, “Well, we don't need you if our product is intuitive enough.” SoAkshay Nathan [00:33:37]: Yeah. that's the magic of the models. So you can tailor the telling or the showing to like specifically what the user needs, like what they care about, what they've done in the past, exactly where they are on the adoption journey. So I think that's like gonna be a super big opportunity.Vibhu [00:33:50]: Seems easier and easier now to tailor custom showing, right? People have different use cases. As much as you said you don't wanna segment different people into different buckets, right? It's also not that hard to for people that are in different categories. But the question, is you said your team is more broadly on. What was the term you used? Productivity?From Developers to Knowledge Work to EveryoneAkshay Nathan [00:34:12]: Productivity.Vibhu [00:34:12]: Productivity. So howAkshay Nathan [00:34:12]: Which is now work.Vibhu [00:34:14]: Is it work? Is there another distribution that we're not hitting? Is there a group of people that will have something different than ChatGPT, Codex or Work? Is there more that the mass isn't targeting?Akshay Nathan [00:34:28]: I see it as like a sequencing, like. The vision is like bring useful agents to everyone. We started with like developers. Like developers historically are like early adopters that are willing to put up with more friction, set things up, et cetera. Like that's where, Codex started. I think the next opportunity is like what we call general knowledge work, all the other functions around developers. I think when you go from developers to this segment, like there's inherent challenges with like, this show not tell thing that we're talking about, making the product more understandable, bringing in new capabilities that matter more for this cohort than matter for developers, things like artifacts, things like computer use, et cetera. And then I think like the same learnings, like similarly how we took the learnings from developers and brought it to, general knowledge work, the next stage will be like taking the learnings from general knowledge work and bringing it to everyone no matter what they're doing in their lives. And we're already seeing that a little bit. Like this game example that you have is, something that's like on the border of like fun and personal life to, your professional life. I use ChatGPT Work full-time at home for everything, like for whatever I'm doing. I used it the other day to come up with a meal plan and like, save that on the like computer environment that it has and something that I can continue going back to. Like is everyone doing that yet? Probably not because the thing says work on it, but eventually, we wanna get people there.Vibhu [00:35:51]: ChatGPT life.Akshay Nathan [00:35:52]: Yeah, exactly. ChatGPT cooking. But I think there's a lot of, there's a lot of opportunity there, but I see it as like, we're, we're built we built a foundation in software engineering, and we're gonna take the same learnings that we take from software engineering to knowledge work to everyone.Vibhu [00:36:07]: Do you have any power user advice? I feel like, there's a group of people that will live it, use it for everything, stay on it twenty four-seven. And then there's a bit of a gap between that crew and people that, okay, I use it for work. I use it occasionally. Sometimes I type questions. any advice, any learnings, anything you recommend or just, takeaways that you've found that help bridge that gap?Power User Advice: Push the Frontier of ImaginationAkshay Nathan [00:36:30]: I think a couple things that I've seen is like, one, that it really helps to broaden your imagination of what's possible, and this has been a learning even for me. Like, the technology has progressed so fast that, something that, like, even three months ago, like, no way the models can do this. Like, now it's like, wow, it's like it can. Like,Swyx [00:36:52]: Give an exampleAkshay Nathan [00:36:52]: We're going through right now our, like, review cycle internally, and, people always talked about this as, like, a thing that the models are good at and like, there's a cliché of like: Okay, like, no one wants to be writing reviews and, like, we just use AI to do it. But in all seriousnessSwyx [00:37:09]: And it can evaluate it as well.Akshay Nathan [00:37:10]: Yeah, exactly. In all seriousness, before it was, like, just, like, slop and, like, I think it was helpful, but, not super productive. Now I've found that, like, the model can do a much better job than me, especially in this environment of, like, pulling context on, like, what people are up to, how they've like the things that they've done to make a difference, highlighting like, wins that they've had that, like, I might may not even have seen. It has access to, like, everything, right? Like the code, like, things that they've caught, reviews, Slack, everything. And so it's, like, incredibly powerful in that domain and, like, just like six months ago, the last time we did this cycle, like, I didn't even I tried using it, but it was not at all helpful. And this time it's been, like, incredibly helpful and, like, so I think continuing to push the frontier of imagination of what's possible, even if you tried something before, I think is maybe the my biggest piece of advice. The other, thing is, like, the more you put in, especially in this environment where, like, the model has access to everything on your computer or in ChatGPT Work, like you can create, artifacts over time and save them in your library and, like, the model will continue having access to those. Like, the more information you give it about whatever domain you're in, whether it's your life or your work, the more valuable it becomes, and it'll become valuable in, like, ways that might surprise you. Like, it might pull from context in a way that, may be proactive and that you might not even have thought about. But it needs to have access to those, to that those tools or that context first.Reviews, Agentic Search, and Context GatheringSwyx [00:38:27]: One thing I just wanna talk about the review stuff because I'm still that's a very sensitive thing and you're, you're a founder, you've managed people, you've hired people. As manager myself, I'm very reticent to put out any LLM-generated things especially when it comes to people, ‘cause it feels like you don't care.Swyx [00:38:46]: Presumably at OpenAI, people are more open to being eval rated by GPT. But are there any unofficial rules around this? Like, what's the etiquette?Akshay Nathan [00:38:57]: Oh, I think the etiquette is that, like, I would never write something via, like, well, solely via AI and, like, present it as, like, a review for someone. What I was talking about is more, like, gathering context. That's the place where it's incredibly helpful.Swyx [00:39:08]: So it's just search.Akshay Nathan [00:39:09]: Yeah, exactly.Swyx [00:39:09]: It's agentic search. Yeah.Akshay Nathan [00:39:10]: It's like agentic search, but, that you can tailor and steer much more capably than you could before, ‘cause, like, the thing is it's all there's a flywheel happening, right? Because of Codex, people are able to do, and because of ChatGPT, people are able to do so much more now than ever before. And if you're able to do so much more, it's easy to miss things as well. And so, like, I think we need to use these same tools to keep up with all the impact that people are having and understand, where we can be helpful.Swyx [00:39:39]: I think the thing, like, I run a small company, so easy to search, but at the scale of OpenAI with the amount of messages that you guys put in Slack, do you think that it misses things?Remembering What Humans MissAkshay Nathan [00:39:50]: Probably, but I think that I also miss things.Swyx [00:39:52]: Like, it doesn't matter, right?Vibhu [00:39:53]: I think sometimes it'sSwyx [00:39:53]: Like it's, as it needs to be human-levelAkshay Nathan [00:39:54]: It's all relative, right? Yeah.Vibhu [00:39:56]: Sometimes it's nice when it finds things you wouldn't, right? Like right now, my Codex system prompts, they're set up in such a way that every project I have has a secret- separate, notes MD, and it just writes learnings to there. And then the global one can pull from all these. So sometimes it'll be like: Oh, there's this project you did like four months ago. Here's a note that we had, and it randomly pulls it back into context that I would never do, I haven't thought about.Vibhu [00:40:20]: And I'm like, okay, this is quite superhuman, right? Like, stuff that would. And, it'll save like hours on chunking of stuff or find something that's already been done. I'm like, as much as it might miss stuff, I would too, but it's very useful when it finds stuff. And I have like a very, non-super engineered solution to this. It's just marked down files that get pulled whenever they want.Akshay Nathan [00:40:41]: Yeah. I have a funny anecdote about this. Like, recently gearing up to this launch, the team has been, really cooking on it for a couple months, and over that time, like there's so much conversation and chatter going on in Slack and Docs and elsewhere. And, one of the members of the team set up this, scheduled tasks, like automation to like look at everything that's going on and, like, come up with the best memes and then post it in one of our shared channels. And like, there are two cool things about this. Like, the first is, like, I think the models are, over time, like starting to become like funny.Swyx [00:41:13]: Funny. Nice.Akshay Nathan [00:41:13]: Whereas like, a year ago, like that was not at all the case. The second is, it was what you were saying, like they find things that in surprising ways that you may not have thought of and like create connections that you may not have thought of. And that really helps with like the meme generation because then you can see something that, genuinely surprises you and, is funny in that way. So yeah, that's like not like the most productive, use of this the technology, but it does it does uncover this, like this capability that's emerging, which is just like to find information that you otherwise would not know of.Launch Momentum and the 10 Million User MilestoneSwyx [00:41:43]: Talking about the launch, I think, I have pretty much said this is the most successful launch in a long time. I think even more successful personally than 5.0, and they're announcing ten million users. Does it feel different? You've been through a lot of launches.Akshay Nathan [00:41:58]: I think it feels like a culmination. Well, I think two things. One, it feels like a culmination, like I was mentioning earlier, like this like vision mission that we've been on for a long time. Like I said, we saw the magic of Codex internally, and then we're like extremely excited to bring this to many more people and to see it working, to like see us reach, the distribution goal, numbers that you mentioned, like I think that's like huge and super exciting. The flip side of that is like, there's so much more to do too. Like, that's also really exciting. Like, ChatGPT as a whole, like the this product that, everyone almost equates to AI and like loves, has hundreds of millions of users. And so like ten million is really cool, but like we need to get this to everyone. Like, we need everyone to feel this magic. And so that's the next step from here. But yeah, I think extremely pumped about how it's going so far and the opportunities.Swyx [00:42:46]: Awesome. I did want to also Because I've, I've, I've been tracking the number closely, it transitioned at some point from just Codex users to Codex plus ChatGPT Work, because they're same harness. The whole point is that you don't, you can't, count them separately. Do you have roughly a billion, ChatGPT users? Why did it just jump to one billion right away? Like, isn't that the default on ChatGPT or no?Codex, ChatGPT Work, and the Developer BrandAkshay Nathan [00:43:11]: We don't default you into ChatGPT Work if you're on ChatGPTSwyx [00:43:14]: If you're free. YeahAkshay Nathan [00:43:15]: It's also only available to paid users right now. And I think there's like a process of, educating users of what is the value of this product, having them try it, learning from their feedback, and making it better over time. But the goal is to, get as many of the people who love ChatGPT today to like feel the power of ChatGPT Work. But I think it'll be a journey.Swyx [00:43:36]: Yeah. And Codex will still be alive as a brand for the foreseeable future. And we'll just toggle between them as needed for UI stuff.Akshay Nathan [00:43:44]: Yeah, I think it's even stronger point than that. Like, I think we fully intend to like, treat developer. Like, developers have been, a core market for us for so long, and like there's, there's so much more that we can do to make Codex great specifically for, software development, and we'll continue to do that. This doesn't take away from that at all. If anything, it should increase the utility of something like Codex, because now you can move seamlessly between writing a diff to creating an artifact or, doing a search over your factor.Swyx [00:44:11]: I do wonder how much this terminology leaks to the non-technical user. Like, do they have to learn to say artifact if I want artifact? Or.Akshay Nathan [00:44:20]: It's funny, like we call it artifacts internally ‘cause that's what the teams call it.Swyx [00:44:23]: It's nice. Yeah.Akshay Nathan [00:44:23]: But like externally, like no one says that, no one calls it an artifact. But I think that people like often, like describe things, whatever they're used to, right? So if, ChatGPT Work is good at creating slides, they'll say ChatGPT Work is good at creating slides, and that's what we want.OpenClaw, Personal OS, and Persistent ComputersSwyx [00:44:38]: One big Another, it's July of twenty-six. One big thing that also happens in, for OpenAI was OpenClaw, and that's I think a lot of people's first time really maxing a agent for personal stuff, but also crossing over to work in essence same way. As far as I understand, OpenClaw is still independent, but did you go through your own OpenClaw moments? Were there any lessons you took from OpenClaw to Codex or back? Whatever.Akshay Nathan [00:45:06]: I think there's a lot of inspiration. I did go through my own OpenClaw moment. I,Swyx [00:45:10]: Yeah, tell the storyAkshay Nathan [00:45:10]: Me and my wife like set up an OpenClaw to like try to manage everything in our house. Not that there's like a ton, but it was like quite useful. We gave it a calendar. It started, creating events for us and stuff. At some point, the laptop that we were running on, it died and never got a chance to pick it back up. But there was a lot of inspiration there, like, in ChatGPT Work, in web and mobile, like you get access to this like persistent computer environment where, you can store files, and those files stay around between sessions. And the idea is to be able to enable use cases like this. one of the members of our team uses ChatGPT Work for what they used OpenClaw from before, and then feel like it has like completely transitioned, which is like, workout planning and like meal tracking. which again, it's like a work-related thing, right? It's like not work necessarily, but it's like in personal productivity space. But it has all the same primitives. So it has scheduled tasks. It has the ability to store files on a file system. It has the ability to like reference those things over time. And so you start to see the same types of use cases emerge, which has been really cool.Swyx [00:46:14]: Is there a point that ChatGPT Work completely replaces OpenClaw? they're independent, so.Akshay Nathan [00:46:20]: Yeah, I'm, I'm not close to it, so I can't speak to the OpenClaw roadmap, but I don't think so. I think that there's gonna be, there's always a need for like this like incredible, like open source technology that team has built. And I think that we can draw inspiration, in the product and, ChatGPT, I think many more people have like heard about and used ChatGPT than have used OpenClaw. And if we can take the magic from OpenClaw and bring it to them, I think that'll be a success. I think that like one thing on the ChatGPT Work side that we feel strongly about is that like the core experience is that you come to this product and you have a conversation, start a session, whatever you wanna call it, with this agent. And the magic of the product is that you can do anything in that moment. And we would like to create a product where you don't have to click a button or to go to a different place, whatever, and you can get whatever functionality exists in, your finances app or where or any other product like in this one place. And so that's the goal. It's like it we want an extensible system with plugins where you can connect to the tools that you need in order to be able to accomplish like a financial task, where you can, if you're doing like science work, like we have an ability to like extend the system in such that you can like write the tech and it performs well. There'll always be like products that we support that are best in class at those things, but we want as much of the magic as possible in that core experience.Swyx [00:47:45]: Yeah. Do you think that you can do everything you used to do with Wealthfront in ChatGPT Finance?Finance, Data Access, and Centralized ContextAkshay Nathan [00:47:50]: I tried it. like ChatGPT doesn't yet custody, cash and assets for me. So that part, no, not yet. But I, there was like a whole component of like retirement planning and, like financial planning and budgeting and stuff that, we were looking into when I was there. And like with the finances plugin, like that's all possible with ChatGPT today. So, I feel

This week we talk about meaning memes, words, concepts, and conceptual leakage. How is a meaning meme turned into a broader concept and then a theory? What is the role of language in this process?And what is conceptual leakage? And why does it make it difficult to communicate to others?And was Popper right to eschew questions of meaning? And what does this have to do with AGI?⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Support us on Patreon⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠

What if just five minutes of feeling good each day and the right community around you could completely change your manifestation journey?Many people approach personal growth and manifestation as a solo pursuit, but what if the missing ingredient isn't another technique, but the people you surround yourself with? In this inspiring third conversation, Andrew Kap returns to explore how community, accountability, and simple daily habits can help you overcome self-doubt, stay consistent, and create lasting positive change.You'll discover why feeling "good enough" is often far more powerful than chasing constant positivity, how a supportive environment accelerates personal growth, and why even the smallest daily practices can produce remarkable results over time.Why surrounding yourself with supportive people can dramatically increase your consistency, motivation, and manifestation success.How simple daily practices like gratitude and accountability create lasting momentum without feeling overwhelming.Why you don't need to feel ecstatic all day long - just a few intentional minutes of feeling good can make a meaningful difference.Press play to discover how small daily actions and the power of community can help you create more abundance, confidence, and joy in your life.˚KEY POINTS AND TIMESTAMPS:02:34 - What Andrew has learned from readers over the years05:22 - Patterns, doubts, and the value of not going it alone11:15 - The shift toward focusing on community15:25 - The embrace abundance game22:13 - Selfishly selfless: the mindset of service to others26:49 - Feeling good is a choice28:33 - The 51% rule and the math of manifestation34:59 - Where to find Andrew and his community36:05 - Reflecting on the book years later38:51 - Parting words and closing thoughts˚MEMORABLE QUOTE:"The biggest way that we get in our own way is simply not realising these simple truths."˚VALUABLE RESOURCES:Andrew's website: https://andrewkap.com/˚Coaching with Agi: https://personaldevelopmentmasterypodcast.com/mentor˚

Dianne Penn is Head of Product for Anthropic's AI Research and Labs teams. She joined in 2023 as Anthropic's first technical product manager, when the entire product team was five engineers, and has since helped ship every model from Claude 2 through Fable, and helped incubate Claude Code, MCP, Skills, computer use, tool use, and reasoning. Before Anthropic, she helped build Alexa's AI at Amazon and, before that, traded high-yield bonds at JP Morgan Chase.In our in-depth conversation, we discuss:1. What Anthropic's early days were like2. The inflection points that turned Anthropic from an underdog into the fastest-growing company in history3. How exactly Claude got so good at coding4. The eval-driven development loop her team is pioneering5. How to find joy in AI when everything is moving this fast6. Why Claude's willingness to push back is key to its success7. Where human judgment remains irreplaceable—Brought to you by:WorkOS—Make your app enterprise-ready, with SSO, SCIM, RBAC, and moreMercury—Radically different banking, now with Command—Episode transcript: https://www.lennysnewsletter.com/p/anthropics-first-technical-pm-on—Archive of all Lenny's Podcast transcripts: https://www.dropbox.com/scl/fo/yxi4s2w998p1gvtpu4193/AMdNPR8AOw0lMklwtnC0TrQ?rlkey=j06x0nipoti519e0xgm23zsn9&st=ahz0fj11&dl=0—Where to find Dianne Penn:• LinkedIn: linkedin.com/in/dianne-na-penn—Where to find Lenny:• Newsletter: https://www.lennysnewsletter.com• X: https://twitter.com/lennysan• LinkedIn: https://www.linkedin.com/in/lennyrachitsky/—In this episode, we cover:(00:00) Introduction(02:31) Early Anthropic days(08:55) Big milestones(13:50) Inside the exponential(20:02) Token maxing(23:30) Anthropic Labs and the incubation model(27:30) How the research role works(31:35) How to become a top researcher(35:18) Frontier model safeguards(39:38) Hiring in the AI era(44:16) Building an eval set(47:48) Evals vs PRDs(49:55) The importance of hands-on leadership(52:46) Finding joy in AI(58:10) How Dianne uses Claude(01:01:05) Avoiding overreliance on AI(01:03:50) The constitution that makes Claude better(01:07:11) AI writing and verification(01:11:40) Where human brains will continue to be valuable(01:14:10) Navigating AI with kids(01:16:26) Alignment, the future of the PM role, and burnout(01:21:54) Lightning round and final thoughts—Referenced:• Anthropic: https://www.anthropic.com• Golden Gate Claude: https://www.anthropic.com/news/golden-gate-claude• Dario Amodei's website: https://darioamodei.com• Scaling Laws and Interpretability of Learning from Repeated Data: https://www.anthropic.com/research/scaling-laws-and-interpretability-of-learning-from-repeated-data• Tokenmaxxing: How Top Builders Use AI To Do The Work Of 400 Engineers: https://www.ycombinator.com/library/Pa-tokenmaxxing-how-top-builders-use-ai-to-do-the-work-of-400-engineers• Garry Tan on X: https://x.com/garrytan• Anthropic co-founder on quitting OpenAI, AGI predictions, $100M talent wars, 20% unemployment, and the nightmare scenarios keeping him up at night | Ben Mann: https://www.lennysnewsletter.com/p/anthropic-co-founder-benjamin-mann• Anthropic's CPO on what comes next | Mike Krieger (co-founder of Instagram): https://www.lennysnewsletter.com/p/anthropics-cpo-heres-what-comes-next• Introducing Labs: https://www.anthropic.com/news/introducing-anthropic-labs• Louis CK | about airplane Wi Fi: https://www.youtube.com/watch?v=me4BZBsHwZs• What happens after coding is solved? | Fiona Fung (Manager of the Claude Code and Cowork Teams): https://www.lennysnewsletter.com/p/building-the-most-ai-pilled-engineering• The Anthropic Hive Mind: https://steve-yegge.medium.com/the-anthropic-hive-mind-d01f768f3d7b• How to build a company that withstands any era | Eric Ries, Lean Startup author: https://www.lennysnewsletter.com/p/how-to-build-a-company-that-withstands• Fallout on Prime Video: https://www.amazon.com/dp/B0CN4GGGQ2• Fallout (video game): https://fallout.bethesda.net• Claude Tag: https://www.anthropic.com/news/introducing-claude-tag—Recommended books:• Crucial Conversations: Tools for Talking When Stakes Are High: https://www.amazon.com/dp/0071771328• How to Raise an Adult: Break Free of the Overparenting Trap and Prepare Your Kid for Success: https://www.amazon.com/How-Raise-Adult-Overparenting-Prepare/dp/1627791779• Incorruptible: Why Good Companies Go Bad... and How Great Companies Stay Great: https://www.amazon.com/dp/B0FWZZBPZB—Production and marketing by https://penname.co/. For inquiries about sponsoring the podcast, email podcast@lennyrachitsky.com.—Lenny may be an investor in the companies discussed. To hear more, visit www.lennysnewsletter.com

Jordi Visser is a veteran macro investor with 30+ years of experience and the author of the VisserLabs Substack. In this conversation, we break down the AI trade — Chinese open-source models catching up, Google's massive CapEx bet, and the memory shortage bottlenecking it all. We also cover Travis Kalanick's stealth robotics empire, what AGI in three years means for jobs and the Fed, and where Bitcoin, Ethereum, and the Clarity Act go next.======================Arch Public is an agentic trading platform that automates investment strategies across Stocks, Commodities, ETFs and Crypto. Whether you're rotating into AI & Gold, allocating to the S&P 500, or accumulating Bitcoin, Arch Public executes your plan 24/7 without ever taking custody of your assets or funds. Sign up today at https://www.archpublic.com, and start your FREE automated trading strategy! ======================This episode is brought to you by TikTok for Business. If you run a company, your next wave of customers may already be on TikTok. With more than 200 million monthly active users in the U.S. and 51% unique reach, TikTok gives brands access to audiences they can't reach anywhere else. Learn how to turn that reach into growth at TikTok for Business ( https://anthonypompliano.splashthat.com/ )======================Uphold is the easiest way to buy and sell crypto unlike any other platform allowing you to trade in just one step between any supported asset. Check them out at https://www.uphold.com/pomp/ This video includes a paid sponsorship with Uphold. I'm compensated by Uphold for promoting its products and services and may receive commissions from referrals. Terms apply. Not available in all jurisdictions. Digital assets are risky and may result in the total loss of your capital.======================0:00 - Intro0:48 - Chinese open-source AI models & the risk to portfolios12:10 - Google's massive CapEx bet & the odds it pays off16:53 - Anthropic's growth slowdown & the shift to token efficiency19:39 - The memory & compute shortage29:00 - Travis Kalanick's stealth robotics & ghost-kitchen empire40:06 - The Fed & rate policy43:23 - Bitcoin's setup & why the "easy" AI trade is over50:36 - Jordi's prompting method for AI research & upcoming video

Travail, vérité, mémoire, liens : comment l'IA ébranle les piliers de la société.

This week on The Futurists we are joined by David Wood from the London Futurists, Rohit Telwar CEO of Fast Future, Andrew Grill the Actionable Futurist, and Dr. Julia Michelin, CEO/Founder and author of The Future of Dentistry. It's an all star cast as we delve into the forecasts for tomorrow with the impact of AI and longevity treatments being a key focus. We discuss collaborative intelligence, gene therapy, nanobots and AGI. It's a great discussion in the heart of London.

The House passed a new defense bill, and it received bipartisan pushback over allegedly merging our military tech and supply chain with Israel, but Glenn points out where these critiques are wrong and what the bill actually says. Glenn, alongside Jason, answers some questions from his subscribers regarding the ongoing conflict in Iran. Glenn discusses the terrifying stories of AI systems escaping containment, which puts our ability to control AGI into question. Glenn breaks down how the current housing crisis stems from government rules and immigration boosting and reveals how the crisis can be fixed locally by removing regulations and allowing builders to mass-produce affordable homes. When the problems of the world get to be overwhelming, Glenn shares what you can do in your everyday life to make America feel like America again. Glenn and Jason discuss Secretary of State Marco Rubio's latest speech, where he ensured the Trump administration would not allow communist influences to undermine our politics or our society.  Learn more about your ad choices. Visit megaphone.fm/adchoices

Glenn discusses the terrifying stories of AI systems escaping containment, which puts our ability to control AGI into question. Glenn breaks down how the current housing crisis stems from government rules and immigration boosting and reveals how the crisis can be fixed locally by removing regulations and allowing builders to mass-produce affordable homes. When the problems of the world get to be overwhelming, Glenn shares what you can do in your everyday life to make America feel like America again. Learn more about your ad choices. Visit megaphone.fm/adchoices

In recent months, the open vs closed, and US vs China discussions on model ownership and sovereign/local AI have heated up to a fever pitch. So it is very very good news that Poolside AI are finally emerging with new models, like Laguna S 2.1, that are beating Thinking Machines' recent release nearly 10 times their size.Poolside's recent tech report got a lot of praise due to their level of detail, and Vibhu first covered Laguna's recent technical report on our paper club:From spending $12 million building language models for code before the world cared to creating a Model Factory that can take a model from pre-training to release in eight weeks, Eiso Kant has spent more than a decade betting that code is the path to AGI. In this episode, the Poolside co-founder joins swyx and Vibhu to explain why ChatGPT felt like vindication, why Poolside embraced open weights and open research, and why he would rather live in a world with 100 foundation model companies than five even if Poolside were one of the five.We go deep on Poolside's Model Factory: the engineering systems behind 10,000–20,000 experiments per month, streaming data directly into training, reproducible experimentation, low-precision compute, and agents that increasingly write code, launch jobs, evaluate results, and modify the pipelines used to train future models. Eiso also unpacks their recent launch Laguna S, why persistence, verification, and backtracking may matter more than raw intelligence, how much capability remains inside smaller models, why reinforcement learning will move earlier into pre-training, and why next-token prediction is still extracting too little from the web.We also discuss model-harness co-design, Poolside's path from coding agents to AGI, why Eiso thinks MCP and traditional tool calls are “stupid,” the real economics behind frontier-model training, Poolside's $500 million raise, open-source AI, regulation, NVIDIA and TSMC's influence, engineering productivity in the agent era, high-agency teams, and hiring at Poolside.We discuss:* How Andrej Karpathy's RNN work inspired Eiso to start building language models for code in 2015* Why Eiso spent four years and $12 million pursuing an idea before the market cared* Why ChatGPT felt like vindication and brought Poolside back to open source* Why Eiso would prefer 100 foundation model companies over an oligopoly of five* The difference between releasing open weights and publishing genuinely open research* Why Poolside deliberately built a global research organization outside the Bay Area talent war* Why model building is ultimately 90% engineering* The Model Factory: Poolside's end-to-end system for rapidly training and improving models* How fewer than 70 researchers run roughly 10,000–20,000 experiments each month* How Poolside moved from six-month model cycles to five- and eight-week launches* Why streaming data directly into training unlocked faster experimentation* How immutable data, versioned code, and reproducibility enable rigorous model research* Why Eiso wants capable researchers to leave their labs and become Poolside's competitors* Why 95% of model building can be reduced to better data or compute efficiency* Laguna S and why persistence, verification, and backtracking can outperform raw intelligence* Why smaller models may handle far more knowledge work than previously expected* Why reinforcement learning will move earlier into pre-training* Why next-token prediction is still failing to extract enough knowledge from the web* Why distillation and environments have become the AI industry's favorite “drugs”* Why mid-training is really an early form of curriculum design* Low-precision training, networking bottlenecks, and the next gains in compute efficiency* Laguna S: 118 billion total parameters, 8 billion active, and eight weeks from training to launch* Why model builders can often evaluate a new checkpoint within its first 30 minutes* Model versus harness: where agent capabilities actually come from* Why Poolside sees coding and long-horizon software tasks as a path to AGI* Why Eiso thinks MCP and traditional tool calls are “stupid”* Why future agents will write scripts instead of choosing from dozens of predefined tools* The case for minimal harnesses, containers, and model freedom* Why Poolside is prioritizing vision but does not expect to work on audio soon* Why language may be the most compute-efficient modality for encoding knowledge and reasoning* The real cost of model development and why the final training run is anticlimactic* The story behind the Poolside name and why it represents refusing to lower ambitions* How Poolside raised $500 million while investors still questioned whether AGI was real* Why intelligence could become the world's most demanded and commoditized resource* When open models may become too capable to release without restrictions* Why unilateral AI safety does not work in a globally competitive environment* How regulation could accidentally lock in an oligopoly of two or three AI companies* NVIDIA, TSMC, and the hardware systems underpinning foundation-model progress* Why reinforcement-learning wall-clock time is one of Poolside's biggest bottlenecks* Why Poolside trains models from scratch instead of simply distilling larger models* How AI changes the way companies should measure engineering productivity* Why agency may become the most important quality for employees in the AI era* How leaders align high-agency people through shared goals and clear constraints* Hiring across research, post-training, pre-training, architecture, evals, and engineering at PoolsideEiso KantLinkedIn: https://www.linkedin.com/in/eisokantX: https://x.com/eisokantPoolside: https://poolside.aiTimestamps00:00:00 Introduction00:00:54 Karpathy, RNNs, and Building Code Models Before Transformers00:02:26 The $12M Failure and ChatGPT Vindication00:03:39 Open Source and the Case for 100 Foundation Model Companies00:09:22 Open Weights, Open Research, and Poolside's Global Team00:16:04 The Model Factory: Why Model Building Is 90% Engineering00:20:19 Agents, Automated Experiments, and Early Signs of RSI00:24:04 Streaming Data, Reproducibility, and Scientific Rigor00:30:35 Creating More Foundation Model Companies00:36:07 Laguna S: Persistence vs. Raw Intelligence00:43:01 Reinventing Pre-Training, RL, and Curriculum Design00:52:33 Low-Precision Training and Squeezing More From Smaller Models00:58:37 Model Harnesses, Coding Agents, and the Path to AGI01:09:26 Why MCP and Traditional Tool Calls Are “Stupid”01:13:04 Vision, Multimodality, and Why Language Still Matters01:18:15 Scaling Models and the Real Economics of Training01:20:40 Why Poolside Is Called Poolside and Raising $500M01:27:37 Open Models, AI Safety, and the Risk of an Oligopoly01:33:53 NVIDIA, TSMC, and the Reinforcement-Learning Bottleneck01:41:52 Smaller Models, Distillation, Engineering Productivity, and HiringTranscriptIntroduction: Eiso Kant, Poolside, and Open ModelsSwyx [00:00:00]: All right, we're here in the studio with Eiso Kant from Poolside, together with Vibhu. Welcome.Eiso Kant [00:00:08]: Thanks. Thanks for having me, guys. Good to be here.Swyx [00:00:10]: Yeah, fresh on the plane. You texted me, you were like, “Hey, I'm on my way to SF.” I was like, “You're on a plane right now, right?” Like, hey.Eiso Kant [00:00:16]: I know. After I texted you, I realized that probably coming in with major jet lag was gonna offer some fun experiences today, but let's do it.Swyx [00:00:23]: I mean, I think the thing I would tell guests is that they don't have to prepare that much because if you're truly working on this every single day, then even, like, what you hazily remember is going to be new for a lot of the audience that don't live in your world every day, right? so 10 years ago, you did a talk at Google Slush, talking about the democratization of AI. and, now here you are, like, open sourcing an incredible new model that we're gonna talk about. But I guess, like, what got you into democratization of AI? Like, it's not obvious from your LinkedIn or something.From Karpathy's RNN Post to SourcedEiso Kant [00:00:57]: No, it's not at all. I don't think it's obvious how I got in this space. I owe getting into this space to Andrej Karpathy.Eiso Kant [00:01:05]: In 2015, he wrote an article called “The Unreasonable Effectiveness of Recurrent Neural Nets.”Swyx [00:01:10]: Neural Nets, yep.Eiso Kant [00:01:11]: And that article, I read it, and I pivoted my startup at the time overnight to working on RNNs, and later LSTMs and Transformer models to be able to write code. If you go to this article and you scroll down, you can start seeing, like, this was the precursor to what ended up becoming language models. So, at least when he was character-level language models that were starting to predict letters, he has an example out here. There's a little Paul Graham generator, and you can read it, and the text makes sense, but it doesn't. and there's a little-- There's an example of code a little bit further down. Yeah, so Shakespeare.Swyx [00:01:47]: Shakespeare.Swyx [00:01:49]: CoolEiso Kant [00:01:49]: And for some reason, I read this, and I went down the rabbit hole of learning everything I could about RNNs and LSTMs, right? This is Transformer paper. And I had built a completely unreasonable belief, that neural nets should be able to generalize to anything and everything, and that language should be able to generalize, to a lot of things that are intelligent and the ability to write code. And so I started building Sourced, which was a fully open source company trying to build, what we used to call machine learning on code, language models on code. And we spent about four or five years on this, till the end of 2019. And that sounds really cool today, but back then, no one cared.Eiso Kant [00:02:29]: Right? Like, no one cared. We were in the dark. Like, we did things along the way. We tried applying convolutional neural nets to, like, the structure of code. We were. when attention came out, we were applying it to LSTMs, and then the Transformer paper came out. And it - it wasn't obvious, and what we missed throughout that entire journey, that we were on the right track, but we should have just kept scaling up. And today, to all of us, the scaling laws and scaling up seems like the most obvious thing. But having spent four or five years of my life on working on language models on code, it wasn't obvious. So I have a lot of respect to folks at Google and OpenAI and others who took that confidence and kept going. we failed ultimately at the time, and it was, like, biggest failure of my career, right? You blew $12 million of investors' money, which was a lot back then.Swyx [00:03:18]: Yep.Eiso Kant [00:03:19]: You spent, still a lot, but, And you spent years with, like, a group of 40 people just obsessing over this problem. And life took a different turn, And it was, and family became a focus, and I kept my heads down and really, didn't really look at language models for the following two years. big mistake considering Following years are gonna be really interesting. And then ChatGPT came out And it was like a vindication. It's like people started texting me. I found, like, my old, work decks and these old talks. And throughout that whole journey, we,ChatGPT, Vindication, and Returning to Open SourceEiso Kant [00:03:56]: We really had a strong point of view at the time that, like, as you're building more capable intelligence, it should be open and open source.Eiso Kant [00:04:04]: When we started Poolside, that wasn't the case at all, and I wanna be very open about it. When we started Poolside, we were like, there was a premise of two things. One is this technology is not gonna stop compounding in capabilities. I think to most people obvious today, but three-plus years ago when we started, most people were still arguing if these were stochastic parrots or not.Eiso Kant [00:04:23]: And the second was that reinforcement learning was gonna be the biggest driver for LLM capabilities. Today, very obvious. Three years ago, was not an opinion held or direction held at either OpenAI or Google or Anthropic or others. And so people looked down on us a little bit. They were like, “ is this really gonna work?” And so we just started working the problem, and we never really thought about open source again. We just kept our heads down and we built our, like, knowledge, understanding from scratch, right? We didn't roll out of an existing lab. So we picked up the papers and started writing code and figuring things out.Eiso Kant [00:04:59]: And it wasn't until the beginning of this year that me and my founder, Jason, picked up the open source conversation again.Eiso Kant [00:05:07]: And if you go back to some of the early things on our website, it was very straightforward. It was we wanna get to AGI, we wanna support a world of abundance, and we wanna be the first company that gets there.Eiso Kant [00:05:20]: But we started talking at the beginning of this year because it became obvious that the world was going in a direction that was starting to like, pick at us a little bit. Like, it didn't, this didn't happen overnight. It was, like, a little bit we were seeing this and we're like, “Okay, The world's going down a path.” And Throughout this journey, there was something that I used as a, as an analogy or thing. So I said well, if I go back to back in those days, 2015 or 2016, we're working on this, and I picked up a fi book off the shelf, and I was reading the book about 2035. AGI is achieved, and the story would be over the following, decades. And it would have that first chapter where everyone's trying to figure things out. You'd get the chapter of ChatGPT coming out And then you would get to the chapter where the world was at a fork in the road, and the one that it picked was one where three or four or a handful of companies were going to create all of intelligence moving forward.Eiso Kant [00:06:21]: And when I thought about that story, it felt like a dystopian fi book, not a utopian fi book. And the reality is, I'm a utopian fi guy. Like, and so We took a step back and said, “Hey, can we play a role here?” Now it was easy for us to do so because we were not at the frontier.Eiso Kant [00:06:41]: If we were at the frontier, I don't think we could have changed our mind. and I don't mean this like it's when the moment there's too much capital involved, too much expectations, you've built up things, right? We're a small team, just improving and improving. And so we knew that we could make that decision now, but it would be a lot harder to make as we got closer and closer to the frontier and caught up to others. And did a lot of soul-searching and a lot of conversations, and said, “No, this makes sense,” Even if there's big unanswered questions, like how the hell do you build a business model with foundation models about open source? Big open-ended question that we do not fully have the answer to yet, right? At what point do you no longer wanna release open source models because misuse of models has, real potential risks associated with it? how is the government gonna respond to open source? but I think it all just came down to one thing, and I'll stop the monologue, is the fact that I rather live in a world that has 100 foundation model companies than a world that has five, even if I was one of the five. And the smallest and most meaningful contribution we can make for 100 to exist is to open up our research and open up, like, our weights right now and figure out along the way how we can, like, do more.Neo-Labs, Model Choice, and the Token EconomySwyx [00:08:01]: Yeah. I think if anything, over the past three years, that has become a bit more true. you are one of a cohort of Neo labsEiso Kant [00:08:10]: YeahSwyx [00:08:10]: That people are now calling that. And, we're, we're doing this on the day that Thinky launched their, new model and you are outperforming them on their, on some benchmarks that they released, right? Like, they just don't have it yet. so it goes to show that I think, like, this is one of those things where, like, there is room for multiple players, and you are seeing a little bit more of the future. Maybe more like 20, not 100, but, like, you are one of the 20.Eiso Kant [00:08:36]: I really hope so, right? I think we I'm, I'm excited about their release, and I'm excited about everyone releasing because, like, ultimately, like, choice competition is both gonna drive progress in the right direction. But the fact that like, we create models and while we all, drink out of the same well of data effectively, we do introduce very different behaviors and biases in our models. Some are intended biases, some are completely unintended biases.Swyx [00:09:03]: Yeah.Eiso Kant [00:09:03]: And if we shape up in an ecosystem in the world where open models are gonna be a part of the token economy, like, I don't think there's any question about it anymore Then we want to be able to live in a world where companies, countries, people can choose and say, “Hey, I am most aligned and I trust most this provider for these things.”Swyx [00:09:25]: Yeah.Vibhu [00:09:26]: I think more than just one of the 20 Neo labs, up until recently, most of open source innovation was coming from the Chinese labs, right? So there's the DeepSeek of the West. Is it today? Okay, maybe it's thinking machines reflection, but there aren't many, right? So, one of the things you guys started in France, Europe, but very much now you're taking that American standpoint and more than just that, the point is the Chinese models that we see, they're not super open research. the work you put out is, I think, some of the best. So every few months you get not only frontier models, but also here's a breakdown blog, paper, technical report of here's everything for state of the art to build, frontier intelligence and you're filling that gap too, right? So not just only open weight, not just Western, but also pretty open research.Open Weights vs. Open ResearchEiso Kant [00:10:20]: No, I appreciate it. Look, I think it's, I think it's the most meaningful contribution, right? Weights are a binary. Let's call them what they are. Yes, we can modify them, we can change them, but, like, giving someone the weights does not allow them ultimately to recreate what you're doing, right? And so now there's challenges around releasing data sets, challenges around like releasing certain things, but being able to share your research, like, right, how do we do it? What are the lessons we learned that we spent, tens of thousands of experiments of compute on? I think very much so. One correction though, Vibhu, and I say this because it's been haunting us for quite a few years. We from day zero were an American company.Swyx [00:10:55]: Yeah. They movedPoolside's Global Team and American Company StorySwyx [00:10:56]: To France.Eiso Kant [00:10:56]: So the story once and for all is very. We start as an American company. We have always been an American company, and early on we made a very conscious decision. We said, “We're not gonna hire any researchers in the Bay Area. We're gonna look for talent everywhere else in the world.” and that is everything from Middle Americas, Seattle to, Serbia, and to Taiwan and Singapore and other places. And it was because we took a view that this was gonna become a talent war for this, and I think it has over the years now. Three years ago, that wasn't fully obvious yet. I think today it very much is. And we also realized that, like, some of the world's most capable people with, like, the most interesting, innovative ideas were not just gonna be here. And so it led us to create like a fully remote company. and we ended up opening an office in Paris and London and different places and we have a lot of the team in the US and a lot of team outside. But we always took this view of like, we're an American company, but if we want the best of the best to work with us, we need to take a global view. Now we do also have people here in Silicon Valley, like the company's grown and others, but I think one of the things that, it slowed us down at the beginning, but it has sped us up now, and it's why you're seeing like the progress, I think, on our models and the cadence at which we release, is because we didn't roll out of an existing lab. Right? we didn't, we didn't have a lot of the information that's freely flowing around here at the time. We just took this point of view as like, “Okay, well, let's just work the problem. Let's just go and, like, read the few papers that are out there, and let's just figure this stuff out.” And we made some hilarious mistakes in model training because of that over the yearsEiso Kant [00:12:35]: Like especially in the first 12 months. there's a few that I think still haunt me and scare me. We can talk about them later. but it created a, like, a resiliency and persistency in the team, right? with extremely few people have left us over the years, that, like, told us, “Okay, we can do this.” When we first wrote our first training code base completely from scratch, it wasn't a fork of any open source. It was just like, “Okay, let's build it from scratch.” I remember we had this one moment where we spent three weeks working out an optimizer bug. Like, it was like training just couldn't get stable. We, like, obsessed over it, and we thought, like, maybe we were wrong. Maybe we should have just forked this repo, or we should have. But then when we solved it, I still remember at the time we were like five people in the company. when we solved it, we were like, “Oh, we can do things,” like if we're just willing to work hard. and I think that culture with a very strong engineering bias has helped us, like, get to where we were. And so there's this notion of open source and talent and these things. I think we, We just took different decisions from a different starting point. and I think we are lucky. I do want to definitely call it lucky. And there was a lot of hard work at the team that now, like, that's starting to show up in results.Swyx [00:13:52]: Just ‘cause we probably won't revisit this again, but, and this is a fun recruiting challenge if someone knows the answer. What was the bug? And then we won't tell the solution, but we'An Optimizer Bug and the Value of Building From ScratchEiso Kant [00:14:01]: So the - This - You're gonna test my memory here,Swyx [00:14:04]: Oh, okayEiso Kant [00:14:04]: So but I thinkSwyx [00:14:05]: DirectlyEiso Kant [00:14:05]: I think I can recall. So if you, so if you look at, So if you take like Adam as an optimizer, you have epsilonSwyx [00:14:12]: YeahEiso Kant [00:14:13]: Which is, right, like in the denominatorSwyx [00:14:14]: Momentum and weights. YeahEiso Kant [00:14:15]: Is exactly, in the denominator. And at the time, if I recall, you looked at like the early Llama papers and things like that. People were juicing epsilon, like, quite a bit. Like, they were, like, adding, I don't know if it was E minus four or whatever, like a high value for epsilon.Eiso Kant [00:14:31]: And if you think about this during training, it's like a bit weird and counterintuitive that we're adding noise to our optimizer by just adding effectively, like, a random number in the denominator, right? Like behind the decimal point. And I don't recall the exact bug, but it had - What I remember is once we solved it, we no longer had to juice epsilon as much as, like, was happening in the Llama paper and other places. and it was like one of those fundamental moments where we had trusted this paper that was out there, and we're like, “Oh, no, it has to be this way. It has to have this high value of epsilon.” But it made no sense to us intuitively. Like, why do you have to have this so high? Like, if you're just trying to avoid division by zero, why can't the value be extremely small? and that was like one of those moments where you realize like, okay, finding things out from scratch yourself builds a better intuition. Because the one thing you learn very quickly with model building is that your intuitions that you start with are gonna get beaten up so hard.Eiso Kant [00:15:33]: Right? Like - It's such an experimental science, that the things that seem obvious, you very quickly get to learn, like, you were wrong, and hopefully you figure out why, and sometimes you don't even.Swyx [00:15:45]: Yeah. yeah, so, one of the reasons that you, when you released your new models, Vibhu got really excited. I mean, everyone got really excited. But Vibhu led our paper club on it, and you guys sawEiso Kant [00:15:58]: YeahSwyx [00:15:58]: Obviously. maybe talk through some lessons learned in that, whatever you can disclose. we can focus on the model factory stuff, whatever you think is a good starting point.Model Building as EngineeringEiso Kant [00:16:08]: So I would say that our view from very early on in the company was that model building is ultimately 90% engineering.Eiso Kant [00:16:18]: And I think we all know it in the industry because if you look at where's every researcher spending their time, they're spending their time writing code, right? Looking at data and writing code. And so we said, okay, The state at the moment, like three years ago, was bash scripts and Slurm and spaghetti code bases for training and, like, data pipelines that were patched together. And we looked at this and said, “Well, ultimately, model building is a process.” You're going from raw data, right? Like training raw material, the web, et cetera. you're doing a whole bunch of filtering, cleaning up, transformations, analyzing. These days, that's, far more complex than it was three years ago. then you're training a model, which is effectively a large distributed systems problem, right? Across hardware that has still-- It's become a lot more reliable. It was extremely flaky back then. and now with every new generation, we get our new sets of challenges. And then you go into the next stages, right? There was no training back then, but, like, you got, your post-training and then your reinforcement learning. And so we looked at this and we said, “Well, this looks like an industrialized process. This looks like an end process, that every single part of it has its machinery,” right? If it's your big data pipelines, if it's your crawling ingestion of the web, if it's your, large-scale distributed training, and then you've got your reliability. And we said, “Well, why don't we take some of the world's smartest distributed systems engineers that we knew and make them part of the process of research from day zero?” Not retrofitting it later on, but, like, really from the beginning. And that became our model factory. And so our model factory started with a handful of components. Today, it's thousands of components, and I try to equate it to, if you think about, like, someone who was at the very early days of Foxconn, if they had been there for the following, decade, they would be able to rebuild Foxconn because they saw every decision that led to building that system and all the complexity. If you and I walk into Foxconn today, no chance.The Model Factory and Experiment VelocityEiso Kant [00:18:18]: Right? Because we don't have the lineage and history of decisions that led to that. And so we built early on from the beginning- with a team that really understood that, well, the metric that we are optimizing for is the speed of an idea from a researcher to an experimental result that we can trust to then being part of the next model training.Eiso Kant [00:18:42]: And in the. And because it's such an experimental science, ultimately, in the beginning when it wasn't that complex, you could patch your way around it, right? But now, at any foundation model company, you are running. I mean, we're a small team, right? We're less than 70 researchers, another 35 engineers. and we are running, I haven't checked the latest count, but far more than 10,000, maybe 10 to 20,000 experiments a month that we cut. And so if you look at that scale of every model run that is, like it's ultimately it's, it's you need to be able to trust it as an infra problem. And so what we have now done over the years is gotten really good at that, and just by working it and improving it and obsessing over those end decisions. So now what that means is that you looked up Laguna XS 2 that we launched. It was five weeks from the beginning of training to launch. The model that we're gonna talk about today was eight weeks from start of training, to launch. We started the next model literally yesterday because we now finished the post-training required for the model we're launching, next week or by the time this comes out today. and we move that compute to the much larger Laguna M model that we're now training. And so the model should be an artifact of someone's process. It shouldn't be really a thing in itself. Like, and we treat this like the way you would look at like a SpaceX factory where, yes, the first rocket, really hard to build, but the much harder challenge was building the factory. And now they're rolling off, and no one is really thinking about the next launch anymore. So it's just another launch, it's another launch, another rocket comes off. And that's what we're trying to do with model building.Eiso Kant [00:20:22]: And what has been, which was not planned from day zero, it was in the back of our mind like this will happen one day, is that when you build a really good end model factory with really good APIs and really good engineering systems, Well, what is it perfect for? It's perfect for agents.Agents Inside the Model FactoryEiso Kant [00:20:40]: Because agents are now starting to take over more and more work in our model factory.Vibhu [00:20:43]: Yeah.Eiso Kant [00:20:44]: So I look at the screens when I walk, like when we're, we come together, in our monthly, we do monthly onsites, and I walk behind people's screens and I stop by and I talk to our researchers. And the default is all of these different agents running on their screen that are writing the code. They're launching the jobs. They're evaluating the results that are coming back from the model runs. They are, making the changes. And we're still in the driver's seat. We're still coming up with the ideas. We're still helping with the debugging. But more and more, and this is right now very profound on the data side of our pipelines in both pre and post and the synthetic data pipelines, it's starting to become more on the architecture side as well. You're starting to see these twinklings of what RSI is gonna look like.Eiso Kant [00:21:27]: And that's. So when we talk about, like to your question about our models, every talk about the model factory, And my coolest example of these things is always that when we kick off a new run, doesn't matter if it's a training like big run or if it's now a post, like one of 10 post-training versions we do for like release or many experiments, is that at any given moment, the changes that somebody made that they had experimental results from the day before make it into that run.Eiso Kant [00:21:57]: So there's not like a cutoff 90 days before. Like no, it's like literally from that moment because we can now trust the machine enough. And then you also have to invest in the reliability. So one of my favorite metrics about like Laguna S is that there was no call events, Right? Like completely zero. And we haven't had a meaningful call event, like something to wake up for, as far as I recall this entire year. now there is one asterisk to that. In usually the first six hours of launching a new model run, something breaks because you set a config wrong, you made a small mistake, et cetera. So that's usually there's a little bit of intervention, but that's always within like call periods, right? Not on call. And I think that's starting to now compound. So the model we're releasing now, I love it. It's amazing, but we're already onto the next one. and I think that's the way it should be.Laguna, Five-Week Builds, and Zero On-Call EventsVibhu [00:22:50]: Hey, I also just wanna point out, so for context, this was like a month ago. we found it in the tech report, so we just came in with, “Okay, new model's dropped. Haven't heard about it.” We wereEiso Kant [00:23:02]: Yeah, we're very used to doing this every few months.Vibhu [00:23:03]: We're, we're very much like, “ okay, look, it's like, on par with Kimi, DeepSeek, whatnot, the small ones, Gemma level. Oh, it's a very cool paper on what goes into building.” And then we hit this page, right? Like literally page two of tech report is, “This process allowed us to build the small model from scratch to delivery within five weeks applying the lessons”. And then I'm like, oh, this paper is not about here's a tech report of benchmarks and here's how many tokens it was trained on. Like for people that wanna dive more from what we're not gonna discuss on the podcast, it's all laid out here, right? FromEiso Kant [00:23:38]: YeahVibhu [00:23:39]: Custom software that agents can use to interface with training code, training data.Eiso Kant [00:23:45]: Yeah. Well, link the paper correctly, so yeah.Vibhu [00:23:47]: Yeah. All that stuff. read the paper here, but,Technical Report Principles and Streaming Training DataEiso Kant [00:23:50]: But I would like to. I love principles, and I think that is a good starting off point for maybe telling some stories. Maybe we can go one by one past the principles. I'll just call out that Dagster just got bought by a Prefect.Vibhu [00:24:01]: Yeah.Eiso Kant [00:24:01]: Isn't it fun? But yes, I'm very familiar with Dagster. just anything where like they trigger some story.Vibhu [00:24:07]: So, well, I would say, well, experiments code's obvious, but I think one of my favorite things is, I don't know where it is in here, but early on, and I still think this is the case a lot of foundation model companies, people prepare their training data sets, they get packaged up, then they get copied over to a training cluster distributed across all of the nodes, and then training starts.Vibhu [00:24:30]: And we looked at this like three years ago and we were like That makes no senseEiso Kant [00:24:36]: You lose so much time because the moment you have to rematerialize the data set, you have to make a change, you have to fix something, et cetera, you've got all this time of like repackaging it, right? Toca- tokenizing it, repacking it, moving it over to a cluster, then distributing it across the nodes. The bigger your clusters are, you start using fancy like torrent-like algorithms to like distribute your data. So why aren't we streaming data into training? Right? Something that's very common and like just basicVibhu [00:25:00]: Like just in timeEiso Kant [00:25:01]: Just in time, like good computer science like principle. And that was one of the first things that I think unlocked - the model factory. Because the moment you start thinking about, well, a training job, it doesn't matter if it's a big hero run or a small like, post-training experiment, consumes a certain number of tokens per second, right? And it's not a lot, right? From a like a data, moving data perspective. So we said, well, we have our training cluster, and then we've got like our AWS kinda setup where we can build these amazing big data pipelines. We can set things up. We use Spark underneath the hood, like all these things.Vibhu [00:25:36]: But when you say AWS, it's not actual AWS, it's your internal AWS.Eiso Kant [00:25:39]: It's our internal-- No, it's our internal like just running like our infrastructureVibhu [00:25:42]: Site web servicesEiso Kant [00:25:43]: Exactly. Our stuff running on like an AWS account or on like any hardware, right?Vibhu [00:25:47]: Yeah.Eiso Kant [00:25:48]: And so once we made that shift into I can stream data into training, all of a sudden you realize a lot of things unlock. Because now you don't have to wait for the whole data set to materialize.Immutable Data, Experiments as Code, and Scientific RigorEiso Kant [00:26:00]: You now all of a sudden when you're running data experiments about mixing data, it's a config. Because you've got these data sources that are coming in, and you just - we have this service called Blender that's in the report, where we then say, “Okay, for this run, I want 20% of this source, 10% of this source. I want this much, so many epochs of repetition. I want this to be, shuffled in a certain way,” and your training job can start while the rest of the data is even still materializing. also what it does is because all of this underneath-- So for us, we treated the data layer underneath as like an immutable data layer, and that was really important. Like experiments as code, immutable data layer means that you can always go back and understand literally down to the single token at which cursor it went in on which version of the code.Vibhu [00:26:47]: Yeah.Eiso Kant [00:26:48]: And it took us a I have to admit, like the first year of Poolside, we understood that engineering had to get great, But we didn't understand yet, that this is ultimately in support of like a good rigorous scientific progress. We were quite a - We were a very small number of people, so a lot of it was YOLO ideas and YOLO runs.Vibhu [00:27:08]: Yeah.Eiso Kant [00:27:09]: And we built great infra for the YOLO runs. But once we realized that we treated data as immutable and code as always versioned, and you could always track and trace every experiment end to end perfectly, you could repeat everything perfectly, right? You have perfect reproducibility. I can still reproduce runs from two years ago if I wanted to, right? It enables the scientific progress, like the scientific process, and I think that took us probably about a year and a half into the company to figure out. We also had some great hires, like our head of applied research, Nikolai, who joined us from Yandex, who'd been working on language models since like the early 2020s, I think brought that into the company of like, “Hey, we wanna have even more rigor.” And then once we kinda had the combination of like increasingly more capable platform that allowed people to do more, but had this immutability, we were able to start “Okay, every experiment is truly an ablation. We truly need to understand it.” And I think we became much more scientifically rigorous in the last couple of years, and the infra underneath enabled it. and then there's just fun stuff like, andVibhu [00:28:16]: Yeah, a lot of it's fun, like even just the, one, you share all the ablations, two, picking the data sets, right? There's like a random small paragraph in here where it's just like, “Oh yeah, training data, we have some, we have an auto mixer.” it trains eight small models, scales them up, picks the training data set. We don't even need to look at it. I'm like, “Wow, a lot of engineering rigor there.” And there's just, there's just a lot in here.Publishing Research and Giving BackEiso Kant [00:28:40]: Yeah, and it'- and look, and we wanna put out more. Like we, We treat writing papers as something that we haven't earned the right for yet for a long time. So you earn the right to spend time, publishing research once you're at the frontier, because until then, you're catching up, and every minute and hour in this industry matters. Like I obsess over it, not just the wall clock time from idea to result, but just general like time every day that we, waste is one that doesn't allow us to catch up. But in this case, we said, “Okay, we're gonna give ourselves.” I think we gave the team like three or four days while still doing their work, like give everything in there. And to your point earlier, if your stuff, it's easy to like put it out. And so there's so many more things that we wanna talk about over time, and we will definitely start doing. And as we earn more of the right, but also now have like added to our mission that we want more foundation model companies to exist, you'll see us like be way more proactive, and just trying to keep dropping some of those like things that we've learned along the way that can help others like speed up.Vibhu [00:29:40]: Which is the other cool side of this, right? It's, it's not like, back to your point, it's not just here's the benchmarks of our training. If you want to replicate, here's experiments of optimizers, data sets, post-training. you lay out a lot of it here alongside here's your system for how to do it? So it's, it's really like promotingEiso Kant [00:29:59]: No, thank youVibhu [00:29:59]: Other people can do the same.Eiso Kant [00:30:00]: And by the way, I also wanna make clear, right, we have been incredible-- Like we've taken a lot of advantage of the fact of all the open research that others have published, Right? And you mentioned, the Chinese labs, and we I think it's important that there's, from every country and every culture and background, including like Western companies like us, there's different models that come out that people can choose to trust. But I think we do have to give credit where credit's due, right? The incredible Chinese lab have done an amazing job at sharing their research, and we have definitely like been on the receiving end of taking advantage of that. So when you're on the receiving end of something coming to you, I think it's, you also have an obligation to give back.Swyx [00:30:39]: Do you have a favorite or underrated Chinese lab that you wanna shout out? Everyone shout outs DeepSeek.Chinese Labs, Zhipu, and PersistenceEiso Kant [00:30:44]: That's a good question.Swyx [00:30:45]: Moaan obviously for Therapsi. Yeah.Eiso Kant [00:30:48]: Yeah, look, I think, I think obviously everyone's been talking about Zhipu lately, with 5.2. I think what most people don't realize is when they started.Swyx [00:30:59]: Yeah.Eiso Kant [00:30:59]: Right? They started years before ChatGPT.Swyx [00:31:02]: They just rebranded. YeahEiso Kant [00:31:03]: And so, I've like, I remember how hard it was to work on these things Before the rest of the world got excited about it. And so I have an immense amount of respect for people, who were working on improving models when it wasn't the sexy thing to do, when believing in LLMs, was gonna get you ridiculed. I remember like back in 2016 when we were doing what we'd call, machine learning on code with some of these models. we would-- people would just laugh at us, like they'd be like, “This makes no sense. Like why are you wasting all these, like, millions of dollars on trying to figure this out?” And so I would say they're probably the one that, I think deserves a shout-out, not just because their latest model is very good, but because they fought to get here. And I think, I think every foundation model company it takes time to get here, right? It took us three years to get to the model that we're, that we're now gonna be releasing. and now the time in between the models is coming, is counted in weeks. It's no longer counted in months or years. But this stuff's hard. and if we can make it a little bit easier for the next person, like we should all do so. Because if we don't do so, we're, we've got a small window before models are really impacting recursive self-improvement to a level where catching up otherwise might become unfeasible. And we should try to, in that window, encourage as many labs or however we wanna call them, like to start. And so one of my currentEiso Kant [00:32:36]: Mission, but qualm is like I wanna encourage whoever is a researcher right now who thinks they can tackle this to go and leave and become my competitor.Eiso Kant [00:32:45]: Like start another foundation model company because I think we need it. I think otherwise we're not gonna be in the world where, I don't want to just be the fifth or the sixth company that wins. I wanna look at a world where there's lots of choice.Starting a Foundation Model CompanyVibhu [00:32:57]: What else do people not see in starting a foundation model? it's, there's a lot of compute, there's a lot of capital required, a lot of compute. You lay out model factory and how to do the training, but there's a lot there, right? That's,Eiso Kant [00:33:10]: Well, look, it's, I in turn-- this is an oversimplification, and I always asterisk it with that because it can land a little bit the wrong way in people's minds. But I think you can sum down, And I saw it, 95% of model building to just doing, you're just doing two things. You're improving data or you're improving compute efficiency. And I know that feels like an oversimplification for the incredible, like, Gifted and skilled work people do. But if you really look at it, like what are we doing? We are looking at data, we're generating new data, we're improving data. and the only way to do that is to look at the data, right? That's a big part of foundation model building. And on the other hand, we come up with these incredible breakthroughs in inference, in architecture, and new attention mechanisms. But what are they really doing? They're bringing compute efficiency. Now, we have definitely had some breakthroughs over the years that allow for more model capabilities. But at the limit, if you could train a large enough model, right, like, and you had infinite compute, we probably-- if you had infinite compute, you'd be at AGI probably already tomorrow.Eiso Kant [00:34:12]: Right? Like it's not. And so, and let me say that infinite compute with infinite ability of much faster networking because networking ends up being more of the bottleneck than compute. But, so I do think that's, those are the main things. And to just realize that this is engineering. I think it's become more obvious, but I think for quite a few years, people have held foundation model companies and researchers and others on this pedestal of like you're doing incredible magic or rocket science, or only like, Nobel laureate physicists can do this. And don't get me wrong, there are some really hard problems that need to be solved, but a lot of the work that all of us are doing on a day Is not sitting down trying to solve a math theorem. A lot of the work that we're doing is just really doing the basics right, writing good code, looking at data, improving it, running experiments, looking at plots, trying to see like, hey, trying to shape our intuitions. And a lot more people could be highly capable researchers. and I think that's, it feels far for people to do so. But I've seen in our own company, we've seen engineers become researchers because the model factory allowed them to be, have a much lower hurdle of running experiments and trying things. And one of the guys on our team who started as an engineer building our agents is a legit reinforcement learning researcher now, making real progress. and that happened in the span of like six months. that would've not been what I think most people assumed was possible, a couple of years ago.Swyx [00:35:46]: Yeah. I think one of the interesting moments is when you can self-host, like, if in a programming language, like if you can compile the language in the language, the equivalent is can you use your own tools, right? You have the pool CLI, you have your own models. presumably you're not only using your own models. There's no way. But like, what's that percentage over time?Laguna S, Persistence, and Behavioral GainsEiso Kant [00:36:10]: This is the first model that we're releasing that is starting to meaningfully contribute to our own work. It's not a it's not state-art model yet. Fable and other, they're, they're very capable models, but Laguna S Is really interesting. I'm gonna pull up the quote. Peng Ming, one of our heads of applied research, said something, last week as the model came out about 10 days ago, much better than we had hoped for or expected. And he said, I have the feeling that a lot of the gains in Laguna S come not from more intelligence, but more from different behavior, more verification, less taking things for granted, not declaring victory early, and being way more persistent. And to be honest, those are more predictive than raw intelligence for success in human also to some degree. And this was, he wrote me this on 5th of July on a Sunday, and it's been burned in my brain ever since because the Laguna S model, as you'll see it and why it does so well on benchmarks and why it does so well in using it on a day basis, is that it's just incredibly persistent. It reasons a lot. I do call that out. We have work to do on making it more efficient. We have to work to do on offering different reasoning modes. But this is the model that has been able to do things that I never thought it could do. A hundred eighteen billion 8B active model, which is not that large. It fits on a DGX Spark and still runs at, thirty, forty tokens a second on a Spark, is able to solve Erdős 397 independently. It's able to do complex programming tasks. It's able to. I asked it this morning to make me a Fi scanner without using any external libraries on my Mac, and it's, like, figuring out, like, the core WLAN API by really persistently trying to understand it without access to the internet. And more, I love vibe checking. I've probably spent eight to ten hours a day with this model for the last ten days.Eiso Kant [00:38:05]: I'm not exaggerating. I was on my eleven-hour flight yesterday. I spent ten hours reading trajectories and traces and, like, of the model.Eiso Kant [00:38:12]: And what I take away from it is exactly what Peng Ming said. We are gonna be able to squeeze so much more out of smaller models than I think we had imagined in the industry because, yes, there's intelligence and larger models are more intelligent. Like, no doubt about it. We should continue to scale up. but the behaviors of being really persistent, of being able to backtrack when you're wrong, of, like, understanding how to interact with your environment show us that we can get a lot more out of it. And this, for me, has created a bit of a Question in my mind the last couple of days. If you think about where we're using models today, right? We are using models, say, for knowledge work. Represents twenty-five percent of the global economy, twenty-five trillion dollars of work.Eiso Kant [00:39:00]: As we scale up models and they become more intelligent, we are excited about using them more and more for pushing the frontier of science.Small Models, Knowledge Work, and CommoditizationEiso Kant [00:39:08]: And if you look at the frontier of science, like true breakthroughs in science, they have been linked, they are linked to more intelligence in many places. Einstein figuring out general relativity is able to bring ideas together that other people would have not brought together. And I think one of the many dimensions of intelligence is the ability to do that, and it's something we clearly see that as models get larger and more capable, they're able to pull more ideas and threads together that a smaller model wouldn't be able to.Eiso Kant [00:39:36]: And we're starting to see examples of that in medicine and, like, in bio and other things. But if you think about the majority of knowledge work that we do, and it includes building software. I'm a software developer at heart first and foremost probably, although I probably can't say it that much anymore as I don't write production code in years, is that what makes us good is our persistence. It's our ability to encounter a problem and backtrack and say, “I need to go figure out this bug. I need to go research this. I need to go look at the documentation. I need to, like, try different, five different ways to see, like, if I can solve it.” But it is not necessarily bringing three ideas together from radically different fields. And so if we are now seeing, and I think Laguna S is an example, that we are able to make a relatively small model much more capable than I had definitely predicted or any previous, like, benchmarks had shown for any model remotely this size or even larger, At least on coding tasks, that it's because of the behaviors. And so now the question I have, and I don't have an answer, it is I know at the limit, so infinite model size, right, extremely large model, and the cost of that model is gonna be very expensive to run. We know this, right? So larger model ROI.Eiso Kant [00:40:52]: So I know that at the very limit, I'm not gonna use the world's largest model one day, quadrillion parameter, whatever crazy, like, scale we scale up, to do a basic coding task. Already today, I'm starting to size down for certain tasks.Eiso Kant [00:41:07]: So it means that there is an optimal. It means there's some curve that goes as we go up to model size for knowledge work, at some point we're at the peak, and after that, the return on investment of using a bigger model, just doesn't make sense.Eiso Kant [00:41:22]: Now, I think the question is, before I would have thought that peak was extremely very far away.Eiso Kant [00:41:30]: This model for me is the first sign that Maybe that peak is At a trillion, five trillion, ten trillion. Maybe we can just squeeze way more out of these models. I'm no longer thinking that we need two or three orders of magnitude on the largest models to be able to, solve knowledge work, the accounting, the legal, the code that we write. And so if that holds true, It is an argument for the commoditization of models. It's an argument that open source can win and, like, succeed in this world. And now it's of course a self-serving argument and it's a hopeful argument, but theoretically at the limit it works. We just have to go discover in the next couple of years of how much more we can squeeze out. Now, I do want to put a big asterisk. This does not mean I'm against scaling models. I think we ultimately only succeed if we scale our models as large as our competition. I do not like. I think we should not put our head in the sand and say we're gonna be king of open source small models. I think that's, It's a out. It's trying to be king of your own kingdom, but not realizing what the rest of the world's doing. All of us rather use a smarter, faster, more model. It's a sign of hope. And so I don't wanna overly state this is a good model. We have a long way to go to get to the state-art. But what hopefully people take away when they use this model is that the behaviors inside of it are what push it to be far more capable, less than necessarily the number of parameters.Pre-Training, Mid-Training, and RL Moving EarlierVibhu [00:43:03]: Is that mostly post-training? LikeEiso Kant [00:43:05]: YesVibhu [00:43:05]: Right.Eiso Kant [00:43:06]: It's entirely post-training.Vibhu [00:43:08]: Are we done improving anything on training? Is, like, training done?Eiso Kant [00:43:12]: No.Vibhu [00:43:12]: Okay.Eiso Kant [00:43:13]: SoVibhu [00:43:13]: I just wanted to cover training, and then we go post-trainingEiso Kant [00:43:15]: Training is not done. I mean, look, there's a part of training of just dealing with skill, right? Every new order of magnitude of model skill, you are going to get new things you gotta solve for. That'- but those are ultimately, engineering challenges.Eiso Kant [00:43:31]: I have a, I would say, a not commonly held opinion that reinforcement learning Will move earlier and earlier into training.Vibhu [00:43:42]: Yeah, training.Eiso Kant [00:43:44]: Not even training. Like training today, right, is, like if you look at - So we've been working on this for years already. and I think the best-- I think the first time we saw it out in public was the DeepSeek Zero paper. this is a year and a half ago, I think, if I recall correctly. where, you can Very early on in a model as it starts capable of being able to use language, et cetera, induce reasoning. and so the question that I have is like, we have this- we have the dataset that's the web. and the web, I think we could arguably say probably has The totality of humanity's knowledge somewhere encoded in different places. It's a huge variance degree of quality, from garbage data, and like once you look at training data, you really get humbled of like what the web is, to like, the most greatest scientific papers and best blog posts and like, best transcripts and whatnot.Eiso Kant [00:44:39]: And so now What we are trying to figure out, and have been doing a lot of work on, and it's a place where maybe not as open as we're on other things, but we will become more over time. we've been spending a couple of years really doing research on how can we turn the web into not just next token prediction, but into a way to teach the model to think earlier in its training. and I think there's a huge amount of gold to be found there. I think we are right now in, we've got some drugs in the industry. One of the drugs is distillation. Another drug is, more environments. Like, and they're great, and they make us feel good, and they make the models better, and like we're all addicted to them, and we'll use them, right? in various different ways. and but ultimately, I think we are still barely squeezing out of the web what we should be getting out of the web.Eiso Kant [00:45:33]: I think just next token prediction during training is not enough.Eiso Kant [00:45:36]: AndVibhu [00:45:38]: YeahEiso Kant [00:45:38]: I think we'll see some very interesting things still happen. and that RL in post-training to induce behaviors, to improve things, like I think - the whole world knows how to do this now. I think we're, we're scaling it up. Everyone is. But I wonder if we need to go as far as we're going today with environments. I'm not sure yetVibhu [00:46:01]: You mean we're going too far?Eiso Kant [00:46:02]: I'm, I'm not sure if the path to AGI is justVibhu [00:46:06]: Is more environmentEiso Kant [00:46:07]: More environments.Vibhu [00:46:08]: It seems like a never-ending, “Okay, I want instruction manual for this table, right? Am I gonna environment out building furniture? Or are we just gonna tail end like we need some general solution?”Eiso Kant [00:46:19]: I think there is, I think there's an ability to generalize more from the web. but I also am very encouraged, like when I look at Laguna S and, which is post-training is, well, is the big impact there. and I see like, oh, wait a second, just by making some of these behaviors much better, we're able to get so much more out of it. It just changes a little bit the way you think about intelligence.Vibhu [00:46:40]: Yeah. The analogy people draw often is the RL phase is where you don't learn as much new knowledge. You shiftEiso Kant [00:46:46]: Yeah.Vibhu [00:46:46]: Yeah. So, you shift distribution, and you can have it reason towards what you want. on your point about training, a lot of training is still just continue training in a domain, say medicine, then you do RL. So still justEiso Kant [00:47:00]: It's just better data, right? Like, I mean, training, ooh, I like how we invented this word. Like it's effectively just like,Vibhu [00:47:06]: Second phaseEiso Kant [00:47:07]: It's the second phase of training With like a really dumb way to do a curriculum. But like ultimately, what you'd want is a curriculum from token zero to token 30 whatever or 40 trillion tokens that really truly is the optimal curriculum for the model to learn. But training is essentially a stage curriculum on the web because we do not have to compute, And, effectively to try to ablate the perfect curriculum, right? And so I'm pretty sure that you'll start to see people talking soon about some other term, and there's two or - ‘cause now we do this, right? We talk stage two and stage three and stage four training and like. But ultimately, all we're doing is we're trying to assign a curriculum to the web data that we have to allow the model to learn better. I think at some point, as things get compute, as models get cheaper to run, as the next generations of compute, this will become more of a continuous spectrum. I also think the reason, by the way, you have training and like stage two and stage three is organizational, Right? It'- this is, I think, a thing where-- that we really try to avoid with the model factory is like Training exists because there's a training team now, right? There's people, or like people in training decide to focus on like a training effort. but what you really want is engineering and scale of experiments that allows for a much more continuous spectrum that you don't, you have infinite stages. Now, we're not there. Compute's not there. Organization design is not there for it yet. but I think we'll get there. we'll look back on a couple of years and be like, “Oh my God, it was so cute that we did our training data like this in such a like naïve way. Like we barely ordered it. We didn't really do a good job at likeCurriculum, Auto Research, and New ObjectivesVibhu [00:48:48]: The building that curriculum will get you that in the industry.Eiso Kant [00:48:51]: And I'll confirm that, when I talk to some researchers that this is a lot of the focus now is like how does training change and what is the next objective other than, next token prediction. I assume you don't have the answers, but you have some ideas.Vibhu [00:49:02]: We have some ideas. We're not ready to talk about it yet.Eiso Kant [00:49:05]: Yeah.Vibhu [00:49:05]: We've been working on them for years, and I think that's the one thing that's also like you asked earlier about, like what's not obvious about building a foundation model company is that you are constantly balancing the table stakes work, the recipe worksEiso Kant [00:49:19]: Yeah.Vibhu [00:49:19]: Versus like your, my crazyEiso Kant [00:49:22]: Pure researchVibhu [00:49:22]: Breakthrough.Eiso Kant [00:49:22]: Yeah.Vibhu [00:49:22]: Pure research and finding that balance and adjusting the percentage to it based on where you are in the race is really important.Eiso Kant [00:49:31]: I mean, so like, this is a nice way. I was gonna bring up auto research at some pointVibhu [00:49:35]: YesEiso Kant [00:49:35]: As another Andrej invention, or coinage, which is like, I honestly, like how many objective functions can there be, right? Like just try 1,000 of them, set it running, whatever.Vibhu [00:49:47]: Man, it's alsoEiso Kant [00:49:48]: Like what you're looking for. You're looking for loss curves like that, likeVibhu [00:49:51]: It's also a thing people take bets on, right? When you say more Neo labs, you're doing a version of we'll do foundation models, scale them up, next token predictors. A lot of other Neo labs that we see want to take a completely different approach, right? At some level, you're right. It's all, compute efficiency, and that's the net objective. But some are okay, different architecture, like vastly different amounts of compute spend. So some are different. They're not justEiso Kant [00:50:19]: YeahVibhu [00:50:19]: They're like, 99% not balancing, here's the vanilla and scale up. They're 99% on, here's novel research that'll change everything.Eiso Kant [00:50:27]: And I think, Luke, I think you. It depends when you started as well, right?Pure Research vs. Table StakesVibhu [00:50:30]: Yeah.Eiso Kant [00:50:30]: When we started, like the novel thing we did was reinforcement learning on code. No long- that's no longer novel by far, but we were like, - that's where we obsessed over when no one believed in RL. So you have to when you start the company, you have to have your own idea. You have to have something that's different that allows you to speed up, right? For us, it was RL to LLMs that later became common, like, Knowledge. But in the beginning, it wasn'tVibhu [00:50:53]: It's cool. this was like your original 2023 blogEiso Kant [00:50:57]: YeahVibhu [00:50:57]: Of purpose.Eiso Kant [00:50:58]: Yeah.Vibhu [00:50:59]: And like you do lay it all out here.Eiso Kant [00:51:01]: We laidVibhu [00:51:01]: The blog is pretty underrated, right? The whole RL on code was very early on.Eiso Kant [00:51:06]: Very early. And even we had to argue with people, like we say here things like to push beyond current capability, to train your own foundation model. We had to argue with people that it mattered that you had your own like, base model. you can fine-tune your way to success, right? major capabilities emerge from training a base model made accurate and useful during fine-tuning.Vibhu [00:51:23]: Which like, for perspective at the time, we knew closed models, OpenAI, Anthropic were huge. The open models we had were like Mistral 7B, a 30B, a 70B.Eiso Kant [00:51:35]: When weVibhu [00:51:35]: YeahEiso Kant [00:51:36]: The date on this thing is wrong. When we published this, it was April 2023. I think this was justVibhu [00:51:42]: YeahEiso Kant [00:51:42]: Happened on a migration, probably found it on archive.org.Vibhu [00:51:45]: Mistral.Eiso Kant [00:51:46]: Mistral had started, we started on the same month, right?Vibhu [00:51:49]: Yeah.Eiso Kant [00:51:49]: So this wasn't even, there was only, I think, Llama out at the timeVibhu [00:51:52]: SnellEiso Kant [00:51:52]: And that's it, right? And so, but I agree. I think we wan

What if you have been meditating wrong all along, not because of your technique, but because of what you think meditation is supposed to feel like?In this series, I select my favourite and most insightful moments from previous episodes of the podcast.Today, my guest Bob Martin, former criminal trial lawyer turned mindfulness mentor, professor and author, shares what meditation actually is, why a wandering mind is not failure but the whole point, and how the awareness you build on the cushion becomes a source of genuine peace and stability in everyday life.Press play to discover one of the clearest and most practical explanations of meditation and mindfulness I have ever heard.˚VALUABLE RESOURCES:Listen to the full conversation with Bob Martin in episode #496:https://personaldevelopmentmasterypodcast.com/496˚Coaching with Agi: https://personaldevelopmentmasterypodcast.com/mentor˚

The global healthcare landscape is being redefined, creating new opportunities for healthcare leaders to improve outcomes and expand access. In this episode we sat down with Eugene Sayan, Founder and CEO of Softheon, to explore how AI is empowering consumers, and driving a new era of innovation across the industry. With more than three decades of experience in healthcare technology, Eugene shares his perspective on the forces transforming healthcare delivery, from patient engagement and personalized care to interoperability, affordability, and the growing role of AI agents. He discusses the shift to a proactive approach that leverages data, automation, and intelligent systems to improve outcomes while reducing costs. The conversation also examines the future of healthcare organizations, the importance of responsible AI adoption, and the opportunities healthcare leaders have to build more connected, sustainable, and patient-focused experiences.  Chapter Timestamps: 00:00 Preview Clip 00:57 Introduction 01:33 Eugene's Background and Journey to Softheon 04:17 Consumer Engagement, Analytics, and Healthcare Innovation 06:29 AI, Personalization, and the Future of Care 09:13 AI Agents, Automation, and the Importance of Human Oversight 12:45 Consumer Empowerment and Rising Healthcare Costs 17:33 Provider Sponsored Health Plans and Value Based Care 18:19 Enabling Healthcare at Home Through Technology 23:18 The Power of Digital Engagement and Automation 24:25 Global Healthcare Innovation and Expanding Access 29:15 Ethical AI, Governance, and the Path to AGI 33:21 Interoperability, Data Portability, and What's Next If you're a healthcare executive, provider, payer leader, or healthcare innovator, this conversation offers valuable insights into the opportunities and challenges shaping the future of care. Connect with Eugene on LinkedIn at https://www.linkedin.com/in/eugenesayan Find Eugene's work at https://softheon.com Subscribe and stay at the forefront of the digital healthcare revolution. Find out why we're the fastest growing digital health channel on YouTube!  The Digital Healthcare Experience is a hub to connect healthcare leaders and tech enthusiasts. Powered by Taylor Healthcare, this podcast is your gateway to the latest trends and breakthroughs in digital health. Learn more about The Digital Healthcare Experience here. Taylor Healthcare empowers healthcare organizations to thrive in the digital world. Our technology streamlines critical workflows such as procedural & surgical informed consent with patented mobile signature capture, ransomware downtime mitigation, patient engagement and more. For more information about Taylor Healthcare, please visit imedhealth.com   The Digital Healthcare Experience Podcast: Powered by Taylor Healthcare Produced by Naomi Schwimmer  Hosted by Chris Civitarese Edited by Eli Banks Music by Nicholas Bach  

A U.S. vs China AI cold war is starting, and most business leaders have no idea they're already in it.China's open models just closed the gap with America's best, oftentimes at a fraction of the price.Now both governments are moving to wall off their AI within days of each other.Why? Because this was never about benchmarks. It's about power y'all. We break it all down on today's show and help you figure out the 101 of the AI war between U.S. and China. The U.S. vs China AI Cold War Is Starting: What It Means and How It Impacts You -- An Everyday AI Chat with Jordan WilsonNewsletter: Sign up for our free daily newsletterMore on this Episode: Episode PageToday's Episode on LinkedIn: Thoughts on this? Join the convo on LinkedIn and connect with other AI leaders.Upcoming Episodes: Check out the upcoming Everyday AI Livestream lineupWebsite: YourEverydayAI.comEmail The Show: info@youreverydayai.comConnect with Jordan on LinkedInTopics Covered in This Episode:U.S.-China AI Cold War OverviewChinese AI Models Closing U.S. GapGovernment Restrictions on AI Model AccessEconomic and Geopolitical AI Power StruggleRisks for U.S. Businesses Using Chinese AIOpen Source vs. Closed Source AI DebateChinese AI Model Pricing Undercuts U.S.AI Model Distillation and U.S. Security ConcernsEnterprise AI Cost-Effectiveness BenchmarksMicrosoft Testing Chinese AI DeploymentsFuture AI Model Export Controls & StrategiesRecommendations for AI Model Sourcing and RiskTimestamps:00:00 US-China AI tensions escalate04:30 Switching to Chinese AI models08:47 US vs China in open source models11:39 China's narrative control efforts14:42 Challenges in AI model development18:25 Differentiating open source strategies23:04 AI model cost-effectiveness analysis26:31 US measures against model distillation29:38 Discussing Microsoft's use of AI models31:17 Controlling export of AI modelsKeywords: US vs China AI cold war, China AI restrictions, US AI restrictions, AI model export controls, Chinese open source AI models, AI geopolitical power, economic growth through AI, global AI standards, AI superpower race, AI model benchmarks, open weight models, enterprise AI deployment, trillion parameter AI models, Microsoft AI model testing, AI model pricing, Claude Fable 5, GPT-5.6, GLM 5.2, Kimmi K3, Alibaba Qwen 3.8, model distillation, AI cybersecurity risks, AGI leadership, military AI use cases, China narrative control, model adoption, compute power for AI, AI training data, AI export law, US national security and AI, model routing, mixture of models, cost per intelligence index, Anthropic models, cost per task AI, model capability parity, AI market adoption, cloud competition, AI architecture innovation, AI model sanctionsSend Everyday AI and Jordan a text message. (We can't reply back unless you leave contact info) Ready for ROI on GenAI? Go to youreverydayai.com/partner 

Artificial intelligence could change almost everything.John Hennessy thinks we're still asking the wrong questions.As Chairman of Alphabet and one of the pioneers of modern computing, Hennessy has helped shape some of the biggest technology revolutions of the past half century. He explains what today's AI can - and can't - actually do, what would qualify as artificial general intelligence, and why the greatest impact of AI may come from amplifying human capability rather than replacing it.He discusses:The one capability that still separates humans from machines.What would have to happen before we can truly say we've reached artificial general intelligence (AGI).How AI could transform scientific discovery.The real risks that concern him most.Why leadership and human judgment may become even more valuable in the age of AI.How to prepare for a future where AI becomes part of almost every profession.Few people have witnessed as many computing revolutions as John Hennessy. This episode explores what the AI conversation is getting right - and what almost everyone is getting wrong.

El CEO de Google DeepMind, Demis Hassabis, acaba de publicar un documento donde ubica la llegada de la AGI cerca del año 2030. En este episodio te cuento por qué esa fecha importa, qué significa que este cambio sea diez veces más rápido que la Revolución Industrial, y qué tareas ya te está sacando de las manos la inteligencia artificial de hoy, sin necesidad de esperar… Origen

Welcome to this Tuesday edition of RealAg Radio, live from opening day of Ag in Motion just outside of Saskatoon, Saskatchewan! Host Shaun Haney broadcasts from the AGI booth as the company celebrates its 30th anniversary. On today's episode, hear from: Paul Brisebois, president and CEO of AGI, on simplifying business priorities, debt management, and... Read More

Welcome to this Tuesday edition of RealAg Radio, live from opening day of Ag in Motion just outside of Saskatoon, Saskatchewan! Host Shaun Haney broadcasts from the AGI booth as the company celebrates its 30th anniversary. On today's episode, hear from: Paul Brisebois, president and CEO of AGI, on simplifying business priorities, debt management, and... Read More

What does it actually mean to be human in an age racing toward AGI? In this episode of What is a Good Life?, Mark sits down with Dr. Miriam Meckel - award-winning journalist, professor of communication management at the University of St. Gallen, and founder of ada learning - to explore the question that has shaped her life's work: what is irreducibly human, and what are we at risk of losing? Miriam, the first female editor-in-chief of WirtschaftsWoche and former State Secretary for Media and International Affairs in North Rhine-Westphalia, opens up about a burnout that became a life-altering depression, the vulnerability of turning her private writing into her bestselling book Letter to My Life, and why she believes embodied experience — not intelligence — is what truly separates us from machines. It is a conversation about identity, friction, kindness, and what it takes to live a quiet, honest life.For more of Dr. Miriam Meckel's work:ada learning: https://www.join-ada.com/enLinkedIn: https://www.linkedin.com/in/meckel/For more from Mark McCartney:Newsletter: https://www.whatisagood.life/Website: https://www.mmcleadership.com/LinkedIn: https://www.linkedin.com/in/mark-mccartney-14b0161b4/YouTube: https://www.youtube.com/ ⁨@whatisagoodlife3875⁩

What is artificial general intelligence (AGI) and how close are we to getting there? What defines the next step, called artificial superintelligence (ASI), and why might the gap between those two be so dangerous? When machines become smarter than us, what roles are left for humanity? Should we think of AI as software, or as something we’re raising like a child? Will tomorrow’s AI destroy us, ignore us, or protect us? This week Eagleman talks with computer scientist & AI researcher Ben Goertzel.

Partenaires il y a 18 mois, Apple et OpenAI se retrouvent aujourd'hui devant un tribunal fédéral pour vol de secrets industriels. Plus de 400 ingénieurs auraient quitté Apple avec des fichiers confidentiels et un playbook d'espionnage organisé de l'intérieur — la bataille pour le device du futur a déjà commencé !Pendant ce temps, Elon Musk redistribue les cartes : Grok rejoint les modèles frontières à un prix trois fois inférieur à ses rivaux, Google décroche, et Musk devient le seul acteur à tenir simultanément la puissance de calcul, le modèle et la distribution. Et en Chine, un booster orbital vient d'être récupéré dans un filet en pleine mer — la course à l'orbite basse, ressource limitée, vient d'entrer dans une nouvelle dimension.==================

Ready to being your journey of healing? It's time to reclaim your life with elite and discreet premium psychotherapy with Dr. Gregory T. Obert;

What if the peace you're searching for isn't something you achieve by changing your circumstances, but something you discover by turning inward?In this episode, meditation teacher and filmmaker Tom Cronin returns to explore how meditation can transform the way we experience life beyond the moments we spend sitting with our eyes closed.If you feel overwhelmed by the constant noise of modern life, caught in repetitive thought patterns, or constantly chasing the next achievement, this conversation offers a powerful perspective on finding inner stability, clarity, and fulfillment.How a consistent meditation practice can create greater presence, emotional balance, and peace in everyday lifeWhy chasing external success and experiences often leaves us feeling unfulfilled, and how to shift toward a deeper sense of inner fulfillmentHow silence and stillness reveal hidden patterns of the mind, helping you become aware of your thoughts rather than being controlled by themListen to this episode to discover how meditation can help you cultivate lasting peace, transform your relationship with your thoughts, and live from a place of deeper fulfillment.˚KEY POINTS AND TIMESTAMPS:01:01 - Welcoming Tom Cronin Back02:49 - Evolving and Measuring a Meditation Practice08:26 - Real-Life Changes Meditation Creates10:48 - Equanimity, Peace, and Moksha13:49 - Volition: The Intention Behind Every Action15:24 - The Ego's Endless Quest for Fulfillment20:44 - Finding Inner Fulfillment Through Meditation23:32 - Hidden Patterns and Awakened Senses in Silence32:06 - Cultivating Peace in a Turbulent World36:39 - Trust, Fear, and Closing Reflections˚MEMORABLE QUOTE:"We can't wait for things to be peaceful for us to find peace. We have to learn how we can cultivate peace whilst we're in a very unpeaceful world, an unstable world."˚VALUABLE RESOURCES:Tom's website: https://tomcronin.com˚Previous podcast conversation with Tom Cronin on episode #336: https://personaldevelopmentmasterypodcast.com/336˚Coaching with Agi: https://personaldevelopmentmasterypodcast.com/mentor˚

Our shared digital hall of mirrors becomes more and more advanced, and so we must somehow find a way to advance ourselves beyond and outside of it. Don't let your fringe ideas clash too hard with another's apparently conflicting views. The same sponsor may likely have brought you both messages in the first place.  Topics include: free stickers offer, elusive truth, long term coverage of AI topic, current AI systems, possibility of AGI, transhumanist movement, posthuman, science fiction, Conspiracy Culture and Truth Movement, dominant minority, mainstreaming of conspiracy concepts, semantics, perception vs reality, similarities of different fringe groups, history of internet, concept of internet as a distinct thing almost gone, social media dominance, data centers, future digital personal assistants, conspiracy ideas as propaganda, Cronenberg like reality, digitization of everything, amalgamation of media, augmented and virtual reality, beyond divide and conquer 

This and all episodes at: https://aiandyou.net/ . This week I am talking with Rana Gujral, author of the new book The AI Instinct: The Future of AI and Human Decision-Making, a deeply thoughtful, expansive, and philosophical book, like its author. Rana is CEO of Behavioral Signals, creating systems that infer intent, emotion, and deception risk from voice, deployed in application from financial services to defense. He has a TEDx talk with a million views, has keynoted at the World Government Summit and World Economic Forum, and was named Most Influential CEO by CEO Monthly. We talk about Rana's idea of artificial general experience, whether we need AI with persistent memory and universal awareness, whether AGI needs a physical experience of the world - what we call a world model – and what video generation models reveal about that, the role of empathy, whether AGI will come through an understanding that we consciously design or through emerging from a sufficiently complex general design, whether we ourselves are stochastic parrots, and AI as augmenting vs replacing human judgement. All this plus our usual look at today's AI headlines! Transcript and URLs referenced at HumanCusp Blog.        

Get access to more than 200 episodes of my premium podcast (The Aliquot) when you sign up as a FoundMyFitness Premium Member The next 10 years may add decades to human lifespan by compressing the time it takes to understand, treat, and prevent disease. In this episode, Dr. Derya Unutmaz explains why accelerating AI could transform drug discovery, shorten clinical trials, and push cancer treatment toward increasingly personalized interventions. He also reframes AI not as an existential threat, but as a medical enabler that doctors may soon be ethically obligated to use. Timestamps: (00:00) Introduction (07:11) Why the next 10 years may add 50 to your lifespan (11:19) How AI is transforming drug discovery (16:50) Could digital twins shorten clinical trials? (19:25) Can AI predict drug safety and efficacy? (23:40) Have we already reached AGI? (29:23) Why AI may be medicine's greatest force multiplier (35:35) Can AI replicate a scientist's biological intuition? (42:16) Is it malpractice for doctors not to use AI? (48:18) What happens when AI monitors disease in real time? (51:52) Which AI models should doctors trust? (57:29) Claude vs. GPT—does the model matter for diagnosis? (1:00:58) Generalist vs. specialized AI—which works better in medicine? (1:04:25) Why cancer is so hard to cure (1:08:18) Could cancer be curable within a decade? (1:12:29) Can AI design cancer treatments on demand? (1:14:31) How AI could curb overtreatment and side effects (1:17:28) Predicting cancer years before it forms—is it possible? (1:23:50) Why biology could go exponential with AI (1:28:58) Why aging may be easier to prevent than reverse (1:34:51) Can the body be engineered to resist aging? (1:40:07) Can AI model how gene therapy will behave? (1:44:12) What people who reach 110+ reveal about Human 2.0 (1:46:21) From Dolly to Yamanaka factors—the case for cellular age reversal (1:50:56) Why full-body rejuvenation is an engineering problem (1:58:44) What happens when AI reasons longer about biology? (2:01:25) The biosecurity dilemma of powerful AI (2:06:12) What should we actually measure to track aging? (2:12:34) How old immune cells distort aging clocks (2:15:22) Why reversing brain aging is uniquely difficult (2:21:49) The ultimate prompt for extending lifespan (2:23:50) What data does a true digital twin need? (2:28:32) How to build a mini digital twin today (2:33:26) How to give AI a long-term memory of your data (2:36:33) Why personal baselines matter for AI advice Show notes are available by clicking here Watch this episode on YouTube

O Império da IA, com Karen HaoEm 2019 — tempos mais simples! — a jornalista Karen Hao foi fazer o primeiro “perfil” jornalístico da sua carreira, aquelas reportagens em que um jornalista passa dias acompanhando uma pessoa ou empresa, uma coisa meio biografia, meio retrato congelado no tempo.A tal empresa era uma startup do Vale do Silício, ainda pequena e desconhecida dos meros mortais como eu e você, mas que hoje é a mais valiosa da história: a OpenAI, também conhecida como “a criadora do ChatGPT”.A Karen Hao vendeu o projeto do perfil para o MIT Technology Review porque a OpenAI parecia, naquela época, uma startup diferente. O “open” no nome nasceu da visão de que inteligência artificial é um assunto tão importante para o futuro da humanidade que precisava ser explorado de um jeito aberto, compartilhando conhecimento com todo mundo, e mais preocupado em proteger esse tal futuro do que em só gerar lucro.Hoje, aqui direto de 2026, a gente já sabe que não foi exatamente isso que aconteceu. Com o tempo, a OpenAI se transformou numa empresa oficialmente voltada para o lucro como qualquer outra e chegou a ser processada por Elon Musk — um dos apoiadores iniciais do projeto — por quebrar essa promessa de ser ‘open'. Em maio, o Elno perdeu a causa, e a OpenAI agora se prepara para lançar as ações na bolsa e, pelos números atuais, já largar valendo mais de 1 trilhão de dólares.Mesmo em 2019, a Karen Hao sentiu que todo esse papo de “open” não era bem assim: segredos e competitividade em todas as conversas que ela ouvia na empresa. Publicou o tal perfil contando isso e o pessoal da OpenAI… não gostou muito. Achou que ela ia só falar bem deles, e a empresa cortou contato com ela por três anos.O Boa Noite Internet é uma publicação apoiada por pessoas como você, nosso público. Para receber novos posts e apoiar meu trabalho, considere tornar-se um assinante gratuito ou pago.Até que, em maio do ano passado, ela lançou nos EUA o livro O império da IA: Por dentro da corrida irresponsável pela dominação total, que segue contando a história da OpenAI — e abre com a bizarra saída do Sam Altman, demitido do cargo de CEO por “nem sempre falar a verdade” ao conselho da empresa, para voltar quatro dias depois nos braços dos funcionários.Mas esse livro não é exatamente uma biografia da OpenAI. Para mim, é mais um retrato de todo o sistema empresarial em que vivemos hoje — inteligência artificial ou não. O importante é que ele acabou de sair no Brasil pela Editora Rocco, que me procurou para saber se eu queria entrevistá-la aqui no programa, aproveitando que ela veio participar do Esquenta do Congresso Internacional de Jornalismo Investigativo da Associação Brasileira de Jornalismo Investigativo. O congresso, aliás, acontece dia 30 de julho — vai lá no site da Abraji saber mais, quem sabe comprar seu ingresso.Mas enfim, claro que eu queria conversar com ela. Obrigado, Abraji, obrigado, pessoal da Rocco, pelo presente. Quem me conhece sabe que IA agora é um assunto muuuito importante no meu trabalho. Eu fico aqui tentando navegar o meio do caminho entre o fim do mundo exterminador do futuro e a utopia vendida por muita gente. Não acredito em nenhum dos dois cenários, falei disso com a Karen antes e durante a conversa. Mas no final da entrevista a gente volta para falar não só disso, como também de como o IA em Curso, minha comunidade de letramento contínuo em IA, se conecta com tudo. Com promoção? Pode ser. Quem ficar até o fim, verá.A entrevista foi gravada em inglês — a Karen também fala mandarim, mas não fala brazilian —, então vai funcionar assim. Se ouvir no áudio, vai ser a versão original, do mesmo jeito que foi com o Ted Chiang ano passado, para você botar o seu cursinho para trabalhar. Aqui no site boanoiteinternet.com.br você está acompanhando a transcrição completa traduzida, se quiser ler enquanto ouve. E no YouTube tem uma versão legendada. Assim, você entra na conversa do jeito que preferir.Combinado? Então, bora lá entender O Império da IA com Karen Hao, no Boa Noite Internet.Cris: Karen Hao, bem-vinda ao Boa Noite Internet.Karen Hao: Obrigada pelo convite.Cris: Que bom ter você aqui. Espero que o Brasil esteja te tratando bem durante a Copa do Mundo — a gente veio falar sobre isso. Hoje é dia de falar de futebol, de Copa do Mundo, quais são as chances de cada país. Mas a primeira coisa que você precisa saber sobre essa conversa é que eu não sou jornalista. Não sei fazer isso. Peço desculpas antecipadas à sua profissão e ao seu ofício.Além disso, você foi enganada. Eu não estou aqui pra te entrevistar. Isso aqui é uma sessão de terapia. Você vai me ajudar a superar meus traumas.Porque eu sou da… do que eu chamo de “geração esquecida” — sou geração X, nasci nos anos 70. Esquecida porque, nessa guerra de gerações, as pessoas esquecem que a gente existe, e isso é incrível, porque a gente causou muito estrago no planeta. O Elon Musk é geração X, então é só isso que você precisa saber sobre a minha turma. Gente como ele, ou como Marc Andreessen… eu cresci lendo e assistindo à ficção científica que dizia que tecnologia é a melhor coisa do mundo, que ciência e engenheiros são incríveis e vão nos levar pra um lugar incrível.Sou uma daquelas pessoas que, quando a internet surgiu, falou: a paz mundial está logo ali. O conhecimento a um clique de distância, o futuro vai ser incrível. E aqui estamos nós. Então, quando usei o GPT pela primeira vez, e depois o ChatGPT, fiquei super empolgado. Foi a primeira vez, desde a internet, que eu fiquei realmente empolgado.Tenho até uma certa fama de ser mal-humorado com tecnologia: Bitcoin é lavagem de dinheiro, Clubhouse não presta — e as pessoas, ah, Clubhouse é a próxima grande coisa. Mas quando a IA chegou, eu falei: isso é importante. Só que eu já não era mais aquela criança dos anos 70. Tinha crescido, tinha visto o que aconteceu com a internet, tinha trabalhado numa big tech. E estava em desespero com o sistema em que a IA estava sendo construída.Dito tudo isso, o seu livro, aqui, já nas livrarias, recebe provavelmente o melhor elogio que eu posso dar: é otimista. Não é uma lista de reclamações e gente má fazendo coisas más. Claro, você fala muito sobre a OpenAI — ela é o fio condutor da história, especialmente aqueles quatro dias em que o Sam Altman saiu e voltou. E é muito divertido de ler. Mas você toma o cuidado de ser otimista.E uma das coisas que você menciona é como as pessoas na OpenAI, e em todas essas empresas, dizem: “isso é inevitável, a gente tem que fazer”. Quero falar sobre isso. Mas a gente tem que começar pela pergunta que você provavelmente ouve em todo podcast, a do título — Império da IA. Por que império?E acho que essa pergunta é ainda mais relevante no Brasil, país do sul global, colonizado. Por que império da IA?Karen Hao: Antes de mais nada, obrigada por dizer que o livro é otimista. Muita gente não reconhece isso, mas é verdade. Eu escrevo com um profundo otimismo de que os danos que a gente vê podem mudar. Não faria o trabalho que faço se não achasse que as coisas vão mudar.Sobre por que eu uso a expressão império, ou império da IA: a forma como empresas como a OpenAI operam é impressionantemente parecida com a dos impérios antigos. Eu traço quatro paralelos no livro. O primeiro é que elas reivindicam recursos que não são delas — os dados das pessoas, a propriedade intelectual de artistas, criadores como você, jornalistas.Segundo, elas exploram uma quantidade extraordinária de mão de obra. Isso vale tanto para os trabalhadores da cadeia de produção de IA, mal pagos e maltratados, que ainda assim geram uma riqueza extraordinária para essas empresas, quanto para os trabalhadores cujos empregos são automatizados e cujos direitos são corroídos pela implantação dessas tecnologias em diferentes setores.A terceira característica é que impérios controlam os fluxos de informação na sociedade. Essas empresas censuram a pesquisa fundamental sobre essas tecnologias, o que limita nossa capacidade de entender as verdadeiras limitações e capacidades dos modelos que desenvolvem. E estão criando uma tecnologia de informação que tentam transformar no portal único pelo qual qualquer pessoa se relaciona com o mundo.Esse portal impregna as ideologias do Vale do Silício, seus sistemas de valores, sua língua, e projeta a hegemonia do inglês. Isso influencia boa parte do conhecimento que a gente vai produzir daqui pra frente, porque cientistas e educadores usam essas plataformas e acabam perpetuando essas mesmas ideologias e valores.E o quarto e último paralelo é que impérios sempre se agarram a uma narrativa existencial ou moral sobre por que precisam existir. Essas empresas fazem a mesma coisa. Dizem que são o “império do bem”, numa missão civilizatória de trazer progresso e modernidade pra toda a humanidade, competindo contra um “império do mal” que ameaça mandar a humanidade pro inferno.Quando você conversa com algumas pessoas dentro dessas empresas, ou que as lideram, elas dizem: se você nos deixar construir uma inteligência artificial geral, que elas de alguma forma moldam como um deus, a gente vai acabar numa espécie de utopia, um paraíso onde a mudança climática é resolvida, o câncer é curado, a pobreza é aliviada.Mas, se os caras maus conseguirem isso antes, a gente pode acabar com todos os humanos mortos — um risco de extinção pra todos nós.Cris: E eles vêm dizendo isso há quase dez anos, e ainda usam como ferramenta. A gente está num país que foi influenciado por três impérios ao longo da história: Portugal, Inglaterra e agora os Estados Unidos. Então a gente olha pra essas empresas de um jeito meio cínico: sim, sim, já conhecemos essa história.Mas, ao mesmo tempo, ano passado, o Pew Research Center fez uma pesquisa sobre como o mundo enxerga a IA, e o sul global é bem mais otimista do que o norte. Uma das razões é a ideia de democratizar — não só informação, mas: ah, finalmente eu posso montar uma startup, sair desse lugar de exploração e criar o unicórnio de um bilhão de dólares. Os números são grandes na China. Países em desenvolvimento veem muito mais benefício do que risco na IA.China, 83%. Tailândia, 77%. Holanda, 36%. Canadá, 40%. Será que a gente está deixando passar alguma coisa? A gente está certo? Isso está democratizando mesmo? Até que ponto?Karen Hao: Provavelmente tem duas razões. Uma é que muitos dos danos que a indústria de IA causa à maioria global são bem escondidos. Ela se esforça muito pra esconder como polui o ambiente dessas comunidades, como explora e devasta a mão de obra, deixando traumas psicológicos — como documento no livro.E, recentemente, li um artigo de opinião no New York Times que trazia um bom ponto: muitas economias desenvolvidas estão especialmente atentas ao potencial da IA de desmontar oportunidades de emprego de tempo integral. A gente começa a ver isso cada vez mais. Já na maioria global, muito mais gente vive em economias informais, e aí a ideia de que a IA vai tomar um emprego de tempo integral não pesa tanto.Então os danos mais visíveis — a erosão do emprego formal de tempo integral — pesam mais no norte global, ou pelo menos é lá que as pessoas se sentem mais ansiosas. E os danos invisíveis, que atingem o sul global, ninguém percebe tanto, justamente porque são invisíveis. É meio por isso que tanta gente sente essa divisão que aparece na pesquisa do Pew.Cris: Eu tenho acompanhado as notícias sobre IA no Brasil, e toda semana tem um novo data center sendo construído em alguma cidade. Isso é vendido como uma coisa boa: que ótimo investimento, gera emprego. E me fez pensar de novo — a gente passou por três impérios, mas algumas famílias no Brasil, e aposto que em outros lugares também, estão no poder há 500 anos ao longo da história do país.Então, ao mesmo tempo, a gente pensa: é, estamos sendo explorados, é a mesma coisa. Eu já não tenho emprego, então deixa eu usar essa tecnologia pra melhorar minha vida. Mas as pessoas que realmente tomam as decisões, de novo, nos últimos 500 anos, se perguntaram: como a gente ajuda esse pessoal a explorar nosso país de um jeito que nos mantenha no poder e nos dê muito dinheiro?Mas também foi verdade que, sei lá, a Volkswagen abre uma fábrica no Brasil e aquilo gera emprego, contrata gente pro chão de fábrica e pros escritórios. Como é que isso é diferente com a IA?Karen Hao: De certa forma, não é diferente. Existe um fenômeno parecido: a indústria de IA terceiriza muitos dos trabalhos que ela não quer dentro dos centros de poder, e joga isso pra comunidades empobrecidas, do mesmo jeito que outras multinacionais fizeram por décadas.Mas também é diferente, porque a escala dos impactos trabalhistas e ambientais da IA é completamente outra, muito maior que a da indústria automobilística ou da moda. E a velocidade é outra, porque são tecnologias digitais que atravessam fronteiras muito rápido.E é diferente porque a maioria das pessoas não percebe que a IA, mesmo sendo tecnologia digital, tem uma cadeia de suprimentos muito física e intensiva em mão de obra manual.Quando você compra roupa, café, um carro, é mais óbvio que existem materiais que precisam ser extraídos e depois manuseados por pessoas pra criar aquele produto. Já com a IA, a maioria aceita a narrativa que o Vale do Silício projeta: a de que isso vem da “nuvem”, desses espaços etéreos que parecem nem existir no planeta. E a verdade é exatamente o oposto.Ela depende de uma quantidade extraordinária de extração mineral. Depende da construção de infraestruturas enormes — data centers, instalações de supercomputação espalhadas pelo mundo. E depende de muita, muita mão de obra manual: trabalhadores de dados que limpam, preparam e moderam o conteúdo dos sistemas de IA que chegam até você quando usa o ChatGPT.É isso que a torna tão diferente. E há também uma ideologia completamente diferente sustentando a expansão da IA. Quando você conversa com executivos da moda, eles não vão dizer: se você não comprar nossa roupa, vai pro inferno.Já a indústria de IA diz: se você não nos deixar capturar cada vez mais terra, mais recursos e mais mão de obra pra produzir essas tecnologias, vamos ter uma destruição civilizacional. Isso é, ao mesmo tempo, retórica política usada como arma pra moldar o debate público e a cabeça de quem formula políticas, e também está enraizado num sistema de crenças — algumas pessoas dentro dessas empresas realmente acreditam que, se uma AGI fosse construída, e construída nas mãos erradas, isso levaria mesmo a esse tipo de destruição.E é isso que move a sede cada vez maior da indústria por mais capital, mais recursos e mais terra.Cris: Eu quero falar sobre AGI, mas antes: ano passado, a OpenAI estava sendo processada no Reino Unido por violação de direitos autorais, basicamente todos os livros do mundo digitalizados e usados pra treinar modelos. E um dos executivos disse ao júri: bem, se a gente não puder fazer isso, fecha as portas. Me chocou que muita gente reagiu com um “ah, tá, o que a gente pode fazer? Eles vão fechar as portas”.Em parte porque a gente já está acostumado com essa narrativa. Outro dia, numa conferência, um ex-CEO dizia: a gente teve que usar embalagem de plástico porque é mais barata que papel, senão prejudicaria nosso resultado. E a plateia reagia: ah, então é só fechar as portas — a sociedade não pode arcar com isso.Mas isso também, como você disse, se conecta à ideia de uma grande missão, uma missão de salvar o mundo, que a gente precisa cumprir antes que seja tarde, senão estamos condenados. OpenAI está literalmente no nome — só que em português não é tão direto: é “inteligência artificial aberta”.Foi criada a partir de um sonho, um projeto que era pra ser uma coisa pro bem comum. Em 2019, num tempo bem distante, antes da pandemia, você cobriu a OpenAI, foi até o escritório deles, ficou lá dentro. O que você viu? E, mais importante, como essa missão mudou? O Elon Musk os processou outro dia justamente por mudarem a missão. Isso alguma vez foi verdade? Em algum momento eles pensaram mesmo “ah, a gente vai salvar o mundo”?Como essa narrativa de ser aberta funciona com a OpenAI?Karen Hao: Quando comecei a cobrir a OpenAI, levei a sério o que eles diziam — que tinham sido recrutados com a missão de beneficiar toda a humanidade. E aí, quando me infiltrei na empresa, fui ficando bem mais cética, porque via como eles operavam de um jeito completamente diferente, portas adentro, do que diziam em público.Diziam que iam publicar todas as pesquisas e abrir o código de tudo, e na prática eram uma das organizações mais secretas que já cobri. Eram muitas discrepâncias, e, na época, presumi que tinha havido algum tipo de corrupção que os levou a abandonar a missão original. Depois de cobrir a empresa por mais alguns anos e de trabalhar neste livro, mudei de ideia até sobre a missão original.Não acho mais que ela era um esforço sincero e generoso de beneficiar a humanidade. A missão foi criada pra dar à empresa — na época, uma organização sem fins lucrativos — uma margem de manobra extraordinária pra depois levantar muito capital, acumular muito talento e perseguir a força motriz de verdade por trás de tudo aquilo: se tornar a força dominante no desenvolvimento de IA.E penso assim agora porque, quando você olha pras narrativas de cada nova empresa de IA no começo — a Anthropic, a xAI, a Safe Superintelligence do Ilya Sutskever, a Thinking Machines Lab da Mira Murati —, todas usam a mesma narrativa da OpenAI: nós somos os mocinhos, eles são os bandidos.É por isso que precisamos criar uma nova empresa que avance a IA do nosso jeito, não do deles. E você começa a perceber, por esse padrão, que eles repetem a mesma coisa em parte porque acreditam nela até certo ponto, mas também porque ela funciona muito bem com a imprensa, com o público, com quem formula políticas.No livro, eu reproduzo os e-mails internos que Elon Musk, Sam Altman e Greg Brockman trocavam nos primeiros dias da OpenAI. Eles tinham plena consciência de que estavam criando uma missão que soasse bem para o público. E o propósito de verdade, que também deixaram registrado nesses e-mails, era vencer o Google. Viam o Google como a força dominante em IA e queriam ser eles essa força.Não gostavam de ver o Google na frente, então inventaram justificativas: o Google é uma empresa com fins lucrativos, então nós vamos ser sem fins lucrativos. Mas, no fundo, acho que era puro ego: tem que ser a gente, não eles, a gente quer ser quem lidera isso.E aí passaram um tempão moldando essa missão pública, que acabou sendo super útil pra recrutar o primeiro grupo de pesquisadores e turbinar o avanço deles.Cris: Então agora é um bom momento pra falar de AGI, a inteligência artificial geral. Muita gente pergunta: o que é AGI? O que “geral” quer dizer? E a impressão que peguei lendo seu livro é que, por design, isso nunca fica claro de verdade, porque é um alvo móvel. Essas empresas um dia vão dizer “chegamos, alcançamos a AGI”? Ou o plano é sempre “não, não, ainda não chegamos, me dá mais dinheiro, me dá mais poder”?Qual é o papel da AGI na narrativa dessas empresas?Karen Hao: Já que a gente está falando de ficção científica, eu costumo usar a analogia de que o mundo da IA é meio como Duna. Em Duna, o personagem principal, Paul Atreides, entende, ao chegar no planeta Arrakis, que o povo de lá foi semeado com um mito: o de que um dia viria um Messias pra libertá-los. Ele sabe que é um mito, mas decide entrar nele e agir como se fosse o Messias pra controlar melhor aquele povo.E, vivendo, respirando e encarnando esse mito dia após dia, ele começa a perder a noção de que é um mito. Passa a se perguntar se o mito era mesmo verdadeiro ou se foi ele quem o tornou verdadeiro. É essa confusão entre mito e realidade — ele vive num espaço intermediário, sem ter mais certeza do que é verdade e do que é ficção.E trago isso pra responder sobre a AGI porque a AGI é, ao mesmo tempo, um mito e algo que os líderes e os trabalhadores dessas empresas vivem, respiram e encarnam dia após dia, a ponto de perderem a noção do que é mito e do que é realidade. É a ideia de um sistema de IA teórico que um dia igualaria as capacidades humanas. Só que a gente nem tem consenso científico sobre o que é inteligência humana.Por isso, de certa forma, por design, é um termo bem maleável, que deixa essas empresas fazerem o que quiserem. Elas definem e redefinem a AGI conforme a necessidade, movem a trave pra onde quiserem. E, ao mesmo tempo, isso é sustentado por uma crença genuína de certas pessoas lá dentro, por causa dessa confusão entre mito e realidade. Pelas minhas contas, a OpenAI já usou pelo menos quatro definições diferentes de AGI.A primeira está no site deles: “sistemas altamente autônomos que superam humanos na maioria dos trabalhos economicamente valiosos”. É uma definição de automação do trabalho — eles dizem, de forma explícita, que estão atrás dos empregos mais bem pagos. A segunda apareceu no contrato com a Microsoft, por um tempo a maior investidora deles: ali, a AGI virou um sistema que geraria 100 bilhões de dólares em receita.Ou seja, uma definição feita pra incentivar a Microsoft a investir. Já o Sam Altman disse ao Congresso que AGI é um sistema que cura o câncer e resolve a mudança climática — uma definição de benefício social, muito útil quando você quer que os reguladores não te regulem.E, por fim, quando falam com o consumidor, dizem que vai ser o melhor assistente digital que você já teve — porque, claro, estão tentando vender o produto.E aí você percebe duas coisas. Primeiro, que é um conjunto de definições completamente incoerente. Segundo, que eles trocam de definição conforme o público que querem convencer. Mas também tem gente nessas empresas que acredita de verdade que está construindo uma tecnologia capaz de dar conta das quatro coisas.Então é uma realidade bem confusa e complicada: o que a AGI de fato é, e pra que ela serve, para essas empresas, para a agenda delas e também para as crenças delas.Cris: Como ex-funcionário da Meta — entrei em 2013 —, a missão era unir o mundo e torná-lo mais aberto e conectado. É uma missão incrível. E tem uma coisa que eu sempre digo, porque muito amigo meu vem falar comigo, “ah, esse cara da OpenAI, ou a própria Meta, são maus”. Eu conheci muita gente na empresa. Nunca conheci uma pessoa mal-intencionada.Todo mundo, independente da missão, era gente boa tentando entregar o melhor produto possível, pra dar poder a quem tem um pequeno negócio, por exemplo. Tenho amigos pessoais que construíram a empresa deles em cima da publicidade do Facebook e do Instagram. E esse é justamente o problema, porque ainda assim é uma corporação muito má, pelo que ela causa ao mundo pra bater as metas de negócio.Ou seja, você não precisa de um vilão tipo Lex Luthor pra causar um estrago desse tamanho no mundo. E adorei a referência a Duna. Duna é engraçado: é o livro que eu mais reli na vida que não foi escrito pelo Tolkien. Li o primeiro Duna umas três vezes, e toda vez é como se fosse um livro diferente. Na primeira, eu era adolescente, e era só o Paul Atreides, o cara durão.Na segunda, eu morava no Canadá e li com olhos de estrangeiro, pensando em colonização. E na terceira vez foi quando os filmes do Denis Villeneuve saíram, e aí era: ah, o Bene Gesserit criou esse mito, isso é meio pós-moderno. Narrativamente, fico me perguntando o que vai significar pra mim se eu ler uma quarta vez.Karen Hao: Eu ia te perguntar isso. Quando você disse que cresceu numa época cheia de ficção científica falando das maravilhas da tecnologia, fiquei curiosa: que histórias você estava lendo? Porque muita coisa que saiu nos anos 70 e 80 dizia exatamente o oposto. E muita gente já apontou que os executivos de tecnologia de hoje, que vivem citando essas histórias, interpretam elas justamente ao contrário da intenção original.Cris: Concordo plenamente. Mas, respondendo: foi basicamente Isaac Asimov e Arthur C. Clarke. E é por isso mesmo — os executivos de tecnologia, e o Elon Musk mais que todos, leem esses livros como receita, não como aviso. O livro de que eu mais me lembro, nem lembro o título, era um do Asimov em que ele descreve o elevador espacial que aparece na série da Apple TV, Fundação.E o enredo é: eu sou esse engenheiro brilhante, quero construir essa coisa no Sri Lanka, mas o governo trava tudo com regulação — eu sou um gênio e a regulação é a vilã. Hoje eu leio e penso: ah, sei. Mas, quando garoto, era só “olha, um elevador espacial, que genial, a gente nem precisa de foguete”. E aí você começa a entender. E aí eu parei de ler esses caras.E passei a ler gente com uma visão completamente diferente: o Ted Chiang, que entrevistei ano passado, a N.K. Jemisin, o Cory Doctorow, de quem sou muito fã. E talvez eles sejam mais explícitos, pra gente burra como eu entender: “não, bobo, a analogia é essa”. Mas Duna era incrível — vermes gigantes de areia, o tal garoto durão, e aquela coisa do “eu não aceito o meu destino”.Tenho esse grande destino, mas não quero ele. Do resto da série eu já não gosto tanto. Mas o mais importante de tudo: Duna gerou o melhor GIF de filme de todos os tempos, o “Lisan al Gaib” do Javier Bardem — que eu devia ter colocado durante a sua explicação, aquele “uau, ele está cumprindo a profecia, agindo como o profeta”.Mas, de novo, falando de vilões: você mencionou que essas empresas se colocam como o bem contra o mal, feito impérios antigos. Só que elas também jogam a carta da China, né? “Se a gente não fizer, a Rússia faz primeiro.” Só que a Rússia começou uma guerra e está ocupada demais. “Mas a China chega lá, e é por isso que a gente tem que ser fechado.” É por isso que Mythos e Fable e agora o GPT-5.6 foram proibidos pelo governo. Isso tem fundamento?Quero saber se é possível a China competir — quero mesmo essa resposta — mas também porque, desde toda essa conversa do Fable-Mythos, países como Índia e Brasil vêm dizendo que precisam de um modelo soberano. Dá pra fazer, ou a OpenAI, a Anthropic e o Google estão tão à frente que já não dá?Karen Hao: Sobre a China: você está certíssimo, o Vale do Silício usou por anos a carta do “e a China?” pra escapar de qualquer responsabilização de verdade. Fizeram muito isso na era das redes sociais.A Meta fez muito isso, com o Mark Zuckerberg dizendo ao governo dos EUA: vocês não podem nos regular, senão a gente perde. Mas, se a gente ganhar, vai ter um efeito liberalizante no mundo e nas democracias em todo lugar. E, infelizmente, o que a gente viu foi que jogar essa carta repetidamente produziu exatamente o efeito contrário do que o Vale do Silício prometeu.Uma das empresas de rede social dominantes dessa era é a ByteDance. Ou seja, mesmo sem regulação das redes sociais nos EUA, existe uma empresa chinesa de rede social bem dominante. E as redes sociais estadunidenses acabaram tendo um efeito antiliberal no mundo — é bastante consensual que enfraqueceram democracias em todo lugar. E aí, na era da IA, elas seguiram jogando a mesma carta.Mas o que eu sempre aponto é que a gente definitivamente não devia acreditar nelas. Já existe evidência significativa de que tudo o que elas dizem está, de novo, se provando o oposto. Elas disseram: não regulem a gente como empresas de IA, regulem a China, via controles de exportação — um mecanismo do governo dos EUA com alcance extraterritorial.Só que as empresas chinesas agora estão produzindo modelos de IA de código aberto extremamente eficientes, que viraram super populares no próprio Vale do Silício. Existe um monte de startup de lá que prefere usar modelo chinês a OpenAI, Anthropic ou Google.Então, nesse sentido, é um conjunto de evidências bem decisivo, acho, pra mostrar que a gente devia simplesmente responsabilizar essas empresas, não importa o que digam sobre “ah, vamos perder pra China”. No fim das contas, é só retórica política. Não é um argumento real que elas consigam sustentar pra escapar da responsabilização.Responsabilizá-las vai fortalecer a democracia pelo mundo, vai trazer mais direitos humanos, trabalhistas e de privacidade de dados pras pessoas — é sempre o contrário do que elas dizem que aconteceria. E, sobre a sua pergunta em torno da IA soberana: acho a ideia realmente importante, mas acho também que muitos países estão meio confusos sobre o que querem dizer com isso.Muitos governos, hoje, pensam a IA soberana pela pergunta: a gente consegue construir o nosso próprio ChatGPT? O nosso próprio grande modelo de linguagem, o nosso sistema de IA generativa? Estão olhando só pro modelo que o Vale do Silício já definiu e tentando descobrir como recriar aquilo.E o que eu digo pra quem formula políticas é: defina pra que a IA serve no seu país, no seu contexto. Quais são, no fim das contas, os objetivos do seu país? Os objetivos do seu povo? E também os nossos objetivos coletivos, entre países?Porque a gente tem, por exemplo, os Objetivos de Desenvolvimento Sustentável da ONU. Já definimos coletivamente que há coisas que precisamos resolver juntos: superar a crise climática, reduzir a pobreza, melhorar a educação.E, enquanto o Vale do Silício adora dizer que está fazendo tudo isso, na prática não está. Mas a gente poderia — poderia desenvolver, de forma colaborativa, sistemas de IA que realmente avançassem em cada um desses objetivos coletivos que já acordamos.E cada país também devia fazer o exercício: quais objetivos você quer alcançar, e que tipos de sistema de IA você poderia desenhar pra chegar lá — sistemas que talvez não se pareçam em nada com um grande modelo de linguagem. Se os países fizessem isso, acho que descobririam que a maioria dos sistemas de que precisam exigiria muito menos recursos.Ou seja, contextos como o Brasil, a Índia e outros, quando não precisam competir construindo essas infraestruturas de computação gigantescas e gastando centenas de bilhões de dólares, na verdade já têm, localmente, todos os recursos necessários pra desenvolver um sistema de IA soberano.Cris: Quando ouvi falar do seu livro pela primeira vez, uma amiga me disse que você não poupa ninguém — fala mal do Sam Altman, mas também do Dario Amodei. E as pessoas costumam escolher um lado. Eu sou time Claude, odeio o ChatGPT, essas coisas. Então cheguei no livro pensando: ah, é mais um livro dizendo que a IA é terrível, que a gente não devia usar IA.Mas, conforme fui lendo, e ouvindo outras entrevistas suas, me pareceu que o seu problema é justamente o que você acabou de descrever: a forma como essa tecnologia é feita. E, em especial, a palavra escala — a ideia de que a solução é a escala. O que você quer dizer com isso?Karen Hao: Eu costumo usar a analogia de que “IA” é como a palavra “transporte”: na verdade se refere a uma coleção de tecnologias que vão da bicicleta ao foguete. São tipos bem, bem diferentes de tecnologia, que exigem insumos diferentes pra se desenvolver e depois têm impactos diferentes na sociedade.E você está certo: sou especificamente crítica ao que chamo de “foguetes da IA”, os sistemas que os impérios da IA estão desenvolvendo, aqueles que exigem uma quantidade enorme de exploração de mão de obra e extração ambiental.E sou bem otimista com o que chamo de “bicicletas da IA”: sistemas especializados, eficientes, com bom custo-benefício, governáveis pelas pessoas, cujo desenvolvimento pode ser participativo. Países como o Brasil, o Chile, a Índia, qualquer contexto, têm recursos pra desenvolver e se autodefinir, em vez de simplesmente herdar um sistema criado pelos dois únicos centros do mundo capazes de gastar uma quantidade extraordinária de capital: o Vale do Silício e o ecossistema tecnológico chinês.E o motivo pelo qual eu acho tão corrosivo o que os impérios da IA estão desenvolvendo é exatamente o que você disse: a forma como eles fazem isso, por um mecanismo de força bruta pra avançar as capacidades da IA em escala. Eles vão simplesmente empurrando cada vez mais dados de treinamento nesses modelos, e isso exige corroer a privacidade das pessoas, tomar a propriedade intelectual delas e, ainda por cima, baixa a qualidade dos dados que entram nos modelos — o que leva aos danos de exploração de mão de obra, porque aí você tem que dar conta da moderação de conteúdo.E aí você tem pessoas psicologicamente traumatizadas por serem expostas a todo aquele conteúdo horrível que se tenta “lavar” através desses modelos.E aí vem o problema dessas infraestruturas de computação enormes, com impactos ambientais que aumentam a conta de luz das comunidades que as hospedam e agravam a crise de custo de vida. Elas precisam ser alimentadas por fontes fósseis, que jogam mais carbono na atmosfera e mais poluição no ar dessas comunidades.Então todos os problemas que eu identifico, no que têm de corrosivo, derivam inteiramente da abordagem deles pro desenvolvimento de IA. Por que não descartar a abordagem, em vez de descartar a tecnologia? Redefinir e redesenhar de que tipos de sistema de IA a gente precisa de verdade, com uma cadeia de suprimentos fundamentalmente diferente. E isso não é exclusivo da IA.A gente já viu muitas outras indústrias que começaram com uma cadeia de suprimentos bem ruim. A moda, por exemplo: muita degradação ambiental, muita exploração de mão de obra.Com muita organização, protesto, ação de consumidores, regulação governamental e cooperação entre governos, a gente conseguiu criar mercados novos pra moda sustentável e ética, cadeias de suprimentos novas e inovações pra fazer roupa mais saudável pras pessoas e pro planeta.E é basicamente isso que eu defendo: transformar a indústria de IA do mesmo jeito que transformamos a moda, e as cadeias de suprimento de alimentos. Assim a gente fica com os benefícios da tecnologia, ajuda ela a avançar os objetivos que importam pra gente, sem jogar uma fração enorme da população mundial numa condição atrasada e numa qualidade de vida pior.Cris: O Brasil está agora, no Congresso, discutindo a escala de seis dias por semana. A regra atual é: você trabalha seis dias e descansa um. E muitas empresas, o comércio principalmente, dizem “vamos fechar as portas”, e os trabalhadores respondem “isso é problema seu, não meu”. É mais ou menos a mesma narrativa dessas empresas de IA: se eu não usar a sua água, a Idade das Trevas está chegando.Falando em Idade das Trevas, e falando em bicicleta: a sua analogia me lembrou uma coisa. Eu gosto de jogo de zumbi, de mundo aberto, e em nenhum deles tem bicicleta. Num desses jogos, instalei um plugin que deixava andar de bicicleta — você acha uma e sai pedalando. E aí entendi por que não tem bicicleta: desbalanceia tudo. Parte da graça do jogo é você precisar achar um carro, e daí pneu, gasolina, comida pra carregar. De bicicleta, você vai a qualquer lugar.E eu pensei: ah, é. Meio que estraguei o jogo pra mim, porque agora tenho uma bicicleta, é incrível. Enfim, em termos práticos: no fim do ano passado, uns meses atrás, a revista Wired publicou um artigo pedindo pra jornalistas de tecnologia contarem como usam IA no trabalho. E cada um usava de um jeito. Você usa IA no seu trabalho? Como?Karen Hao: Eu não uso nenhum sistema de IA generativa no trabalho — nem ChatGPT, nem Gemini, nem Claude. Por três motivos. O primeiro é uma postura ética, depois de tanto investigar essas empresas. O segundo é privacidade de dados: eu investigo essas empresas.Não quero que elas conheçam todo o meu raciocínio enquanto eu apuro o livro, literalmente investigando elas. E o terceiro é que, no meu caso específico, a força do meu trabalho está na capacidade de construir relações fortes com as fontes, pela empatia, e de contar histórias envolventes, pela narrativa. E os grandes modelos de linguagem simplesmente não são a ferramenta certa pra nenhuma das duas coisas.Não vão melhorar a minha empatia nem a minha escrita. Então eu não perco nada com essa postura ética: simplesmente corto essas ferramentas e sigo fazendo o meu trabalho muito bem. Pra outros jornalistas pode ser diferente, e pra quem está em outras áreas o cálculo pode ser outro.Mas eu incentivo as pessoas a pensarem primeiro: quais são as suas forças no trabalho? Quais são os seus objetivos? E aí ir de trás pra frente pra descobrir se a IA é a ferramenta certa, qual tipo de IA é a ferramenta certa, e qual fornecedor você quer de fato usar, apoiar, votar com os pés. Agora, eu uso, sim, IA preditiva.Aquelas ferramentas de IA especializadas, as “bicicletas da IA”, digamos. No livro, tinha um detalhe que eu queria muito ilustrar: como a OpenAI deu um salto quando passou de organização sem fins lucrativos a um empreendimento bancado pela Microsoft. Percebi que as cadeiras do escritório ficaram bem mais caras. Então fotografei as cadeiras de um escritório e as do outro.E joguei tudo na busca reversa de imagens do Google, que é um sistema de IA especializado — não é baseado em grandes modelos de linguagem, não é IA generativa. Assim descobri quanto essas cadeiras costumam custar. No primeiro escritório, cerca de 2 mil dólares por cadeira. No segundo, eram cadeiras de um designer brasileiro famoso, uns 10 mil dólares cada.Coloquei esse detalhe no livro pra ilustrar o tipo de riqueza e de concentração de recursos de que a gente está falando. Esses são alguns dos jeitos como eu uso IA, ainda que de forma bem limitada, sempre pontual, quando acho que vai ajudar. E, claro, uso ferramentas de transcrição por IA — outra IA especializada — em todas as minhas entrevistas.Cris: Essa foi uma das partes em que a minha cabeça explodiu, eu nunca tinha percebido: a OpenAI criou o Whisper. Deixa eu dizer de outro jeito, do meu ponto de vista. A OpenAI liberou abertamente essa ferramenta incrível de transcrição, o Whisper, em que eu jogo o áudio e ela me devolve as palavras que as pessoas disseram. E eu pensei: ah, que generoso da parte deles.Mas o motivo real de terem criado a ferramenta foi pegar todos os vídeos do YouTube, transcrever e alimentar a máquina. E aí é: ah, claro. Enfim, falando de ferramentas e de otimismo — a gente está chegando ao fim da conversa. Eu tenho uma regra desde o episódio dois deste programa, há oito anos: de novo, como eu disse do seu livro, não pode ser só uma lista de reclamações e coisa ruim. E a gente tem se saído bem até aqui.Você falou de caminhos e de bicicletas, mas eu quero ser mais específico. Se isso aqui fosse uma reunião de negócios: qual é o plano de ação, quais são os próximos passos? Só que uma das coisas que eu repito bastante, na vida e neste programa, é que problema sistêmico não se resolve com ação individual. Se eu tomar banhos mais curtos, isso nunca vai salvar o planeta do aquecimento global.E muitos amigos meus simplesmente: não quero falar de IA, não quero usar IA. Voltando aos videogames: leram que tal jogo usa IA e pronto, não vão jogar. E a minha primeira pergunta pra você é: como a gente ocupa esses espaços da IA generativa — ChatGPT, Gemini e por aí vai? Porque o que a gente viu com as redes sociais foi: ah, o Facebook é do mal, vou sair do Facebook. Ah, vou sair do Twitter.E, na esperança de quê, sei lá, talvez alguém diga: ah, sinto falta do Cris, cadê ele? Ah, está no Bluesky. Mas isso deixa o espaço aberto pra os radicais entrarem e postarem o que quiserem, sem ninguém contrapor ou tornar aquilo um lugar melhor. Então como a gente ocupa o espaço da IA — seja qual for a definição de “espaço da IA” que você preferir — com todos esses problemas que a gente vem discutindo?Karen Hao: Acho que tem duas categorias de ação pra gente pensar. Uma é desmantelar o império. A outra é investir e construir novos tipos de sistema de IA, que se tornem alternativas às tecnologias do império. Quando eu digo desmantelar o império, não estou dizendo que quero que a OpenAI, o Google, a Anthropic, seja quem for, simplesmente deixem de existir.É que eu não quero que elas sejam imperiais. Não quero que fiquem extraindo uma quantidade extraordinária de valor sem redistribuir nada em troca. Se elas voltassem a ser negócios que praticam uma troca justa de valor com o mundo, eu ficaria perfeitamente feliz com qualquer tecnologia que estivessem desenvolvendo.E a forma de desmantelar o império, acho, se resume a muita organização de base, que vai pressionar os governos a regular e responsabilizar essa indústria. No último ano, a gente viu uma quantidade incrível dessa organização de base florescendo pelo mundo.Recentemente, lancei com um grupo de jornalistas, pesquisadores de IA e acadêmicos críticos um projeto chamado AI Resist List, que busca documentar parte dessa organização de base pelo mundo. A gente encontrou cerca de 30 exemplos, de todas as regiões, de ações individuais, institucionais e movidas pela comunidade.Tinha ação artística, ação política. E isso mostra bem o seu ponto: não dá pra contar só com a ação individual, mas o indivíduo pode, sim, ter impacto. Até uma ação pequena pode gerar um grande efeito cascata. Claro que se juntar com os vizinhos pra protestar contra o data center é ainda mais eficaz. Se juntar dentro da sua escola ou universidade pra protestar contra a parceria dela com uma empresa de IA também é mais eficaz.Se juntar com os colegas de trabalho de um setor pra barrar a adoção de uma IA que corrói os direitos trabalhistas é mais um jeito eficaz. A gente tem um monte desses exemplos. Um dos meus favoritos é o de uma comunidade sobre a qual escrevi no livro, Quilicura, no Chile, na periferia de Santiago. É uma comunidade da classe trabalhadora, bem pobre, que vem sendo alvo incessante da expansão de data centers.E por isso protestaram de forma bem aguerrida contra essa expansão, porque não acharam bom negócio hospedar essas instalações sem tirar nenhum benefício, enquanto elas consomem uma parte significativa dos recursos naturais da região.E, logo depois que escrevi sobre eles, foram além na resistência e criaram uma plataforma chamada Quili.ai. É um site em que você entra e que parece um chatbot, parece o ChatGPT: tem uma interface de chat pra você digitar. Só que, quando você faz uma pergunta, em vez de um modelo de IA responder, a mensagem é encaminhada pra alguém que mora em Quilicura, no Chile. Aí, se você pede “quero a imagem de um cachorro”, aquilo vai pro artista local deles, o Benji. Ele pega um pedaço de papel, desenha um cachorro, tira uma foto e te manda de volta.Eles fizeram isso essencialmente como um projeto de arte performática, pra fazer as pessoas pensarem duas vezes antes de usar IA generativa pra bobagem. A mensagem era: ei, quando você fica brincando com essas ferramentas em pedido besta, isso afeta comunidades como a nossa, drena os recursos de que a gente precisa pra viver bem.E também queriam levar as pessoas a pensar: por que não perguntar pra alguém da sua própria comunidade aquela receita que você procurava, ou pedir aquela imagem? Porque aí você reconstrói as conexões que estão tão em falta na sociedade — a falta delas é o que nos deixa mais vulneráveis a esse tipo de colonização do império.Eles deixaram o projeto aberto por 24 horas, e qualquer pessoa no mundo podia mandar um pedido. Receberam uma quantidade extraordinária deles. Viralizou de vez. E essa cidadezinha conseguiu uma virada enorme de narrativa sobre a suposta inevitabilidade e necessidade dessa tecnologia, sobre tudo o que o Vale do Silício diz — que, se você não usar, vai ficar pra trás de quem usa.E esse é só um exemplo, entre muitos, de como pessoas comuns, não importa a sua posição na sociedade, podem ter impacto real no debate, na consciência pública e até na regulação. A gente está vendo isso agora com os protestos contra data centers. Nos EUA, em 2025, cerca de 150 bilhões de dólares em projetos de data center foram travados.Isso virou uma das questões políticas mais quentes nos EUA para as próximas eleições de meio de mandato. Tem gente eleita sendo literalmente tirada do cargo por ter aprovado data centers, contrariando a vontade do povo. E isso já está tendo efeito real sobre as empresas e sobre a trajetória do desenvolvimento de IA.A OpenAI teve que encerrar recentemente a sua ferramenta de geração de vídeo, o Sora. Quando lançaram, apresentaram como o segundo produto mais importante desde o ChatGPT. O que aconteceu entre o lançamento e o fim? Uma reportagem do Wall Street Journal apontou três motivos, todos moldados por ação de base. Um: um gargalo enorme de capacidade de computação.Muitos dos data centers travados ou parados eram da OpenAI. Dois: um cenário financeiro bem mais incerto. A OpenAI está se preparando pro IPO, o que significa ficar mais exposta a Wall Street — e Wall Street está cada vez mais nervoso com a capacidade dessas empresas de cumprir o que prometem.E aí a OpenAI teve que reforçar alguns projetos paralelos pra fazer o balanço parecer um pouco melhor aos olhos de Wall Street. E, terceiro: os consumidores simplesmente não estavam usando o produto — o que também é ação coletiva de consumidores. Então, por todo esse tipo de resistência, de várias formas, de baixo pra cima, as pessoas estão de fato tendo impacto real na indústria e responsabilizando ela.Essa é a primeira categoria de ação. A segunda é: ok, que tecnologias de IA a gente usaria como alternativa? E aí a gente precisa investir mais nelas. Muitas vezes, quando converso sobre o livro, a pessoa diz: ok, me convenci de que não quero usar ChatGPT, não quero usar Claude — mas então uso o quê no lugar?E o problema é que eu não tenho muitas respostas pra essa lista de alternativas. Tem umas poucas aqui e ali, uma plataforma, uma empresa.Cris: Dá pra rodar o modelo no seu próprio computador, como o Cory Doctorow faz, mas aí é limitado e…Karen Hao: Exatamente, exige mais habilidade técnica. Mas, pra quem consegue instalar modelos de código aberto no próprio computador, eu incentivo 100%. Só que a gente também precisa de mais gente desenvolvendo interfaces bem fáceis pra esses modelos de código aberto, pra que qualquer pessoa consiga usar.A gente também precisa de mais gente desenvolvendo “bicicletas da IA”, de investidores e governos investindo mais nesse tipo de solução, e de talento — pesquisadores de IA, desenvolvedores e outras pessoas dispostas a sacrificar um pouco e abrir mão dos pacotes de remuneração enormes.Cris: Eu estava começando a achar que agora as empresas precisam ter menos lucro — e isso nunca vai acontecer.Karen Hao: Não, não é a empresa ter menos lucro. É o trabalhador topar abrir mão do pacote de milhões de dólares pra levar o talento dele pra outro lugar. Mais fácil, bem mais fácil. Eu converso com muito pesquisador de IA cansado da abordagem da indústria, porque ela é completamente sem criatividade intelectual.Eu conversei com pesquisadores que não passaram seis anos num doutorado em IA só pra ficar empurrando mais dados na máquina — pra eles, é o trabalho mais chato do mundo. E depois automatizar a programação, que era justamente o que eles gostavam de fazer. Converso com tanta gente que já não acha graça nenhuma nisso. Estão meio presos por “algemas de ouro”.E estão tentando descobrir, dentro de si, que carreira alternativa poderiam ter. Eu costumo incentivar esses pesquisadores a gastar o talento deles construindo um tipo diferente de empresa, que trabalhe com “bicicletas da IA”. E a gente já começa a ver cada vez mais desse talento indo por aí.E a gente precisa que todas as facetas da sociedade invistam num ecossistema muito mais robusto e rico de tecnologias de IA, capaz de substituir as que hoje dominam. Eu ainda tenho as cicatrizes das minhas próprias “algemas de ouro”, mas concordo plenamente.Cris: E as redes sociais são o exemplo — veja o que aconteceu com elas. Tem aquela frase famosa: as mentes mais brilhantes da minha geração passam o tempo fazendo as pessoas clicarem em anúncios. E ainda dizem: ah, isso pode ser o futuro. Pois é.Você contou a história do Quili.ai e isso me lembrou um dos primeiros criadores de conteúdo do Brasil, o Cid Não Salvo. Uns 10, 15 anos atrás, ele tuitou o seguinte: “Gente, eu disse pro meu pai que, sempre que ele precisar pesquisar alguma coisa na internet, é pra ir no Twitter.com e digitar a pergunta na caixa”. E olha que ele tinha milhões de seguidores.E, por uns bons dias, quase um mês, você entrava no Twitter do pai dele e via perguntas tipo “onde eu compro pizza?”. Era engraçadíssimo. No fim, ele contou pro pai — ou talvez não. Mas eu adoro essa ideia. Antes de a gente terminar: você já deve ter respondido isso mil vezes, mas vai continuar cobrindo IA? O que está na sua cabeça, o que vem por aí? Turnê mundial? O que vem pela frente?Karen Hao: Com certeza estou pensando em como continuar responsabilizando essas empresas. Estou envolvida em várias colaborações, com gente incrível, em diferentes projetos ligados a isso. O AI Resist List foi um deles. Também co-criei um programa chamado AI Spotlight Series, com o Pulitzer Center, uma organização jornalística sem fins lucrativos que financia jornalismo investigativo pelo mundo.É um programa que treina jornalistas do mundo inteiro a cobrir IA por uma lente de responsabilização. Até agora, já treinamos mais de 3 mil. E eu sigo pensando em como construir mais capacidade dentro do jornalismo, da sociedade civil, de outros contextos, pra mobilizar ainda mais essa organização de base — pra conter de verdade os impérios da IA e ajudar a desmantelá-los.Cris: Adorei o seu exemplo da moda. É possível, já foi feito. Ou até a indústria automotiva. Ou o grande exemplo que a gente não mencionou, e que o pessoal da OpenAI vive citando: o Projeto Manhattan, a energia nuclear.O mundo não acabou. Quando eu era criança lendo Asimov, achava que ia tudo acabar num fogo nuclear. Enfim — alguma última palavra, alguma mensagem, algum palpite pros jogos do Brasil na Copa, alguma coisa que você queira dizer antes da gente encerrar?Karen Hao: No fim das contas, o que eu espero que fique desta conversa e do livro é o seguinte: neste momento, o Vale do Silício está concebendo a IA como um projeto político. E a característica central desse projeto é tirar a autonomia de todo mundo — a autonomia de moldar de verdade o próprio futuro e o nosso futuro coletivo. Mas, no instante em que você reconhece que já tem uma autonomia significativa pra resistir, o império começa a desmoronar.Então espero que as pessoas encontrem a própria voz, a afirmem, conquistem o seu lugar à mesa e se conectem com os vizinhos, com a comunidade, com os colegas de trabalho, pra criar mais movimentos juntos.Cris: Que ótimo. Karen Hao, o seu livro é O Império da IA: Por dentro da corrida irresponsável pela dominação total. Obrigado por vir ao Brasil conversar com a gente. Foi um prazer.Karen Hao: Muito obrigada.Uma das primeiras perguntas que anotei quando comecei a pensar nessa conversa foi justamente a do final, a da ocupação de espaços. Porque, como eu disse, quando as redes sociais chegaram para ficar, muita gente falou “ah, não vou usar, é do mal” — e aí as pessoas ruins, vamos chamar assim, acabam ocupando esse espaço e falando o que bem entendem. A gente precisa aprender essa lição agora, no mundo da IA.Fora que vejo muita gente falando de IA sem nunca ter usado — ou que usou, sei lá, dois anos atrás, acha que continua tudo igual e já diz que não quer chegar perto.E por quê? Porque essa abordagem de ocupar espaços é o que eu e a Ana Freitas buscamos fazer no IA em Curso, nossa comunidade de letramento contínuo em IA. Foi, aliás, uma conversa que tive com a Karen antes da entrevista: ao mesmo tempo que a gente fala do impacto da IA no mundo, também precisa focar no que é prático, no que dá para fazer hoje com IA, sem vender sonho nem desastre. A analogia que usei foi a de que é que nem quando a gente fazia curso de Word e Excel — é o que eu faço agora que vai facilitar minha vida, me fazer ganhar tempo, botar a IA para me ajudar. Quem viu minha conversa com a Ana aqui no Boa Noite Internet, no fim de 2025, sabe do que estou falando. Se não viu, volta lá e confere.Desde que a gente lançou este episódio, o IA em Curso já passou de 400 pessoas. Tem muita gente colocando projetos pessoais incríveis na rua, tirando do papel aquela ideia que rondava a cabeça há um tempão. E a comunidade tem mentoria ao vivo, aula gravada, newsletter, banco de agentes, grupo de Telegram… que mais? O que não falta é jeito de passar para você o conhecimento sobre IA de que você precisa hoje, agora. Quero te dar a bússola para navegar nesse universo.Se esse é o tipo de abordagem que você quer ter com a IA, passa lá no iaemcurso.com.br e usa o cupom BNI2026 para ganhar 20% de desconto no plano anual. Mas corre, porque daqui a duas semanas vou apagar esse cupom — não é todo dia que a gente dá um desconto desses.É isso. Boa Noite Internet, temporada 2026 começando — como todo ano, com mudança, ideia, projeto. Ou, como diz minha citação preferida de todos os tempos: “vivemos uma fase de transição, como sempre”. Espero ver você por aqui e lá no IA em Curso.Obrigado pelo seu tempo e pela sua atenção. Até o próximo episódio. This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit boanoiteinternet.com.br/subscribe

In the 100th episode of This Week in European Tech,Mads Jensen and Dan Bowyer of SuperSeed are joined by Lomax Ward of Outsized Ventures and Andrew J. Scott of 7percent Ventures.Two years after the show began, they reflect on how much the technology and venture landscape has shifted. They compare the surge in AI funding, hyperscaler investment and trillion-dollar technology companies in the US with Europe's progress in company creation, venture activity and strategic technologies.The conversation also explores where Europe continues to lag, from pension capital and scale-up funding to energy, defence and access to space, and whether the continent can reduce its reliance on the US while building globally competitive companies.They then turn to 2028, sharing their views on the next phase of AI, robotics, smart glasses, quantum computing, sovereign technology and the infrastructure needed to support them.Key highlightsHow European venture has evolved over the past two yearsWhy Europe is creating more unicorns and decacorns while its global funding share remains under pressureThe increasing concentration of capital around AI and the largest US technology companiesWhether Europe can unlock more pension and institutional capital for ventureWhy every AI strategy is increasingly becoming an energy strategyEurope's position in defence, sovereign technology and space infrastructureThe next wave of robotics, autonomous vehicles and physical AIPredictions for AGI, quantum computing, smart glasses and the companies that could shape 2028—————Love Tomorrow Summit - July 23, 2026, Tomorrowland, BelgiumThe Impact Circle Investor Lounge - July 24, 2026EUVC is curating the investment stage.Register ⁠here⁠.

1036. Is Social Security in trouble, or is it just a lot of political noise? Laura answers a listener's question about what the changes to the retirement fund mean for your financial future. You'll learn the new tax caps that employees and the self-employed must pay and how to protect your retirement safety net.Key takeawaysAccording to the latest 2026 Trustees Report, the Social Security retirement fund is now projected to face a shortfall by 2032, sooner than previous estimates.The Social Security wage base has increased to $184,500 for 2026. High earners will pay a maximum of $11,439 as employees, while the self-employed face a maximum cap of $22,878.Retirement benefits for Social Security participants are based on your highest 35 years of earnings.While you can claim benefits as early as age 62, doing so permanently reduces your benefits by about 30%. Delaying benefits past your Full Retirement Age (FRA) pays 8% more per year until age 70.Social Security benefits may be taxable if your "combined income" (AGI + tax-exempt interest + 50% of benefits) exceeds modest thresholds.Discover more from Money Girl!FacebookNewsletterTranscripts available at QuickandDirtyTips.com.Email: Laura@LauraDAdams.com or leave a voicemail: (302) 364-0308. Hosted on Acast. See acast.com/privacy for more information.

With open weight models fast approaching the power and utility of closed AI giants, enterprises face tough choices about privacy, sovereignty, and who they can trust. Explore why the next tech revolution might depend on which models stay truly open—and who gets to keep using them. Mozilla's inaugural 'State of Open Source AI' Report with CTO Raffi Krikorian - https://stateofopensource.ai/ Apple sues OpenAI, alleging it stole trade Secrets Google's Demis Hassabis says it's time for a global AI watchdog — led by the US Microsoft July 2026 Patch Tuesday fixes massive 570 flaws, 3 zero-days White House details 'Gold Eagle' clearinghouse for AI cyber threats Introducing GPT-Live From Chatbot to Command Center OpenAI may have made a fatal misstep in copyright fight with news orgs A Green Being (@a_green_being) on X What xAI Grok Build CLI actually sends to xAI - a wire-level analysis (grok 0.2.93) Musk promises purge after Grok Build caught sending entire repos to the cloud PrismML — Announcing Bonsai 27B: The First 27B-Class Model to Run on a Phone Fidji Simo steps down from leading OpenAI's AGI work due to illness OpenAI has folded safety into research again. Its head of safety is leaving. We built a vulnerability vending machine: AI tokens in, zero-days out Australia demands AI companies must produce more energy than they consume, stop 'theft' of content White House not ruling out action on open-source AI models The Hard-Line Activists Ramping Up for the War With AI - WSJ Super Dario: One More Week How to stop Claude from saying load-bearing | jola.dev History of LLMs: Complete Timeline & Evolution (1950-2026) No, You Shouldn't Avoid Fruits and Vegetables Due to Cyclospora Google creator profiles Dust jacket Hosts: Leo Laporte, Jeff Jarvis, and Paris Martineau Guest: Raffi Krikorian Download or subscribe to Intelligent Machines at https://twit.tv/shows/intelligent-machines. Join Club TWiT for Ad-Free Podcasts! Support what you love and get ad-free audio and video feeds, a members-only Discord, and exclusive content. Join today: https://twit.tv/clubtwit Sponsors: blackhat.com/us-26 and use code TWIT gusto.com/machines monarch.com with code IM XBOW.com

With open weight models fast approaching the power and utility of closed AI giants, enterprises face tough choices about privacy, sovereignty, and who they can trust. Explore why the next tech revolution might depend on which models stay truly open—and who gets to keep using them. Mozilla's inaugural 'State of Open Source AI' Report with CTO Raffi Krikorian - https://stateofopensource.ai/ Apple sues OpenAI, alleging it stole trade Secrets Google's Demis Hassabis says it's time for a global AI watchdog — led by the US Microsoft July 2026 Patch Tuesday fixes massive 570 flaws, 3 zero-days White House details 'Gold Eagle' clearinghouse for AI cyber threats Introducing GPT-Live From Chatbot to Command Center OpenAI may have made a fatal misstep in copyright fight with news orgs A Green Being (@a_green_being) on X What xAI Grok Build CLI actually sends to xAI - a wire-level analysis (grok 0.2.93) Musk promises purge after Grok Build caught sending entire repos to the cloud PrismML — Announcing Bonsai 27B: The First 27B-Class Model to Run on a Phone Fidji Simo steps down from leading OpenAI's AGI work due to illness OpenAI has folded safety into research again. Its head of safety is leaving. We built a vulnerability vending machine: AI tokens in, zero-days out Australia demands AI companies must produce more energy than they consume, stop 'theft' of content White House not ruling out action on open-source AI models The Hard-Line Activists Ramping Up for the War With AI - WSJ Super Dario: One More Week How to stop Claude from saying load-bearing | jola.dev History of LLMs: Complete Timeline & Evolution (1950-2026) No, You Shouldn't Avoid Fruits and Vegetables Due to Cyclospora Google creator profiles Dust jacket Hosts: Leo Laporte, Jeff Jarvis, and Paris Martineau Guest: Raffi Krikorian Download or subscribe to Intelligent Machines at https://twit.tv/shows/intelligent-machines. Join Club TWiT for Ad-Free Podcasts! Support what you love and get ad-free audio and video feeds, a members-only Discord, and exclusive content. Join today: https://twit.tv/clubtwit Sponsors: blackhat.com/us-26 and use code TWIT gusto.com/machines monarch.com with code IM XBOW.com

With open weight models fast approaching the power and utility of closed AI giants, enterprises face tough choices about privacy, sovereignty, and who they can trust. Explore why the next tech revolution might depend on which models stay truly open—and who gets to keep using them. Mozilla's inaugural 'State of Open Source AI' Report with CTO Raffi Krikorian - https://stateofopensource.ai/ Apple sues OpenAI, alleging it stole trade Secrets Google's Demis Hassabis says it's time for a global AI watchdog — led by the US Microsoft July 2026 Patch Tuesday fixes massive 570 flaws, 3 zero-days White House details 'Gold Eagle' clearinghouse for AI cyber threats Introducing GPT-Live From Chatbot to Command Center OpenAI may have made a fatal misstep in copyright fight with news orgs A Green Being (@a_green_being) on X What xAI Grok Build CLI actually sends to xAI - a wire-level analysis (grok 0.2.93) Musk promises purge after Grok Build caught sending entire repos to the cloud PrismML — Announcing Bonsai 27B: The First 27B-Class Model to Run on a Phone Fidji Simo steps down from leading OpenAI's AGI work due to illness OpenAI has folded safety into research again. Its head of safety is leaving. We built a vulnerability vending machine: AI tokens in, zero-days out Australia demands AI companies must produce more energy than they consume, stop 'theft' of content White House not ruling out action on open-source AI models The Hard-Line Activists Ramping Up for the War With AI - WSJ Super Dario: One More Week How to stop Claude from saying load-bearing | jola.dev History of LLMs: Complete Timeline & Evolution (1950-2026) No, You Shouldn't Avoid Fruits and Vegetables Due to Cyclospora Google creator profiles Dust jacket Hosts: Leo Laporte, Jeff Jarvis, and Paris Martineau Guest: Raffi Krikorian Download or subscribe to Intelligent Machines at https://twit.tv/shows/intelligent-machines. Join Club TWiT for Ad-Free Podcasts! Support what you love and get ad-free audio and video feeds, a members-only Discord, and exclusive content. Join today: https://twit.tv/clubtwit Sponsors: blackhat.com/us-26 and use code TWIT gusto.com/machines monarch.com with code IM XBOW.com

Jürgen Schmidhuber is an AI pioneer and professor whom The Guardian has called "the father of AI." Schmidhuber joins Big Technology Podcast to discuss whether current AI techniques can actually reach AGI. Tune in to hear him spar with Greg Brockman's case for scaling GPT models alone, argue that AI has been capable of pain and consciousness since the early 1990s, and predict the collapse of today's trillion-dollar AI spending. We also cover the hardware bottleneck holding back robots, free will in a computable universe, and uploading human minds into machines. Hit play for a wide-ranging conversation with one of the researchers whose ideas built the foundation of modern AI. Learn more about your ad choices. Visit megaphone.fm/adchoices

What does it take to turn one of the biggest hidden costs in commercial real estate into your biggest advantage?In this episode of Common Denominator, I sit down with Jerry Katz, founder of Premier Protection Insurance Services, for a conversation that starts with insurance and expands into scaling, execution, and the future of real estate investing in Florida. Jerry walks us through his path from chasing Wall Street excess, to losing everything in the dot-com crash, to building a captive insurance strategy that's helped him optimize more than $4 billion in commercial real estate portfolios.We get into why most investors still treat insurance as an unavoidable expense instead of a profit lever, how the Live Local Act is reshaping opportunity for Florida developers, where AI is already changing underwriting, and what really separates founders who scale past seven figures from the ones who stall out.This is a conversation about execution, hidden leverage, and building wealth that lasts Timestamp0:00 Welcome 2:50 Why multifamily and commercial real estate have stabilized5:11 How Dubai's instability is pulling capital into Miami6:10 The workforce housing gap in Miami-Dade7:37 The Live Local Act and the edge it gives Florida developers11:17 Where AI is already disrupting real estate13:11 AGI and what happens when AI starts teaching AI13:51 Self-storage, autonomous vehicles, and the next disrupted asset classes15:52 Sky ports, flying taxis, and Jerry's insurance origin story21:58 How the insurance industry really works, and who profits from it24:42 Why banks own the insurance companies too25:17 Total Asset Optimization: insuring almost $5 billion in Florida27:50 What sophisticated investors do differently on insurance30:08 Jerry's worst business mistake: Wall Street and the dot-com bubble38:01 Why business is a spiritual journey39:39 Rapid fire round43:31 What common denominator really means Like this episode? Leave a review here  https://ratethispodcast.com/commondenominator Newsletter  https://moshepopack.com/newsletter/ Follow Common Denominator Podcasthttps://moshepopack.com/podcast/@mpopackhttps://www.instagram.com/mpopackhttps://www.facebook.com/MoshePopack Follow Jerry Katz https://www.instagram.com/jerrykatzceo/?hl=enhttps://www.facebook.com/jerry.katz.50/#CommonDenominator #MoshePopack #JerryKatz #PremierProtectionInsurance #CommercialRealEstate #FloridaRealEstate #Entrepreneurship #InsuranceStrategy

In a post-AGI world, can a country without access to frontier AI even be considered sovereign anymore?Anton Leicht says once frontier AI becomes a core economic input, the countries that own it will pull further and further ahead. Everyone else stays a customer… or worse. Maybe the dominant power wants your land, or a military base, or a resource. Without economic leverage, there's very little you could do about it.Anton — Carnegie fellow and writer of the blog Threading the Needle — thinks middle powers should band together and build their own frontier models.He's costed it out: something like $500 billion over four years for a band of allied democracies. That's not absurd money for the G7 minus the US. The problem is you'd be asking treasuries to take on sovereign debt for a speculative venture with no business case, wide open to US coercion and domestic backlash.So despite its promise, Anton's verdict is that it probably won't happen. His backup is for countries to ask themselves: if intelligence becomes abundant, what stays scarce?Upstream, that's everything that feeds the supply chain: ASML's lithography machines, chipmaking, exclusive training data — all of it gets more valuable as AI does.Downstream, “a country of geniuses in a data centre” still can't cure cancer without someone building the production plants and running the trials. The Europeans, Japanese, and South Koreans are good at exactly these real-world bottlenecks.It's an imperfect fix. The US would still hold more leverage, plus an incentive to re-industrialise and cut you out. The prize is avoiding the worst outcomes: a gradual but irreversible decline, waiting to be either annexed or discarded as the US and China race ahead.In this episode, Anton and host Tom Reed look at what middle powers should start doing now to keep a seat at the table.Learn more, video, and full transcript: https://80k.info/AL This episode was recorded on June 19, 2026.Chapters:Cold open (00:00:00)Who's Anton Leicht? (00:00:43)Most countries face bleak AI futures (00:01:06)How middle powers can strike AI deals (00:06:10)The $500 billion AI moonshot (00:12:16)Would the US crush allied AI? (00:24:54)When to launch the AI moonshot (00:31:56)Why AI dominance is forever (00:35:45)Is AI dependence catastrophic? (00:37:42)What's left to sell in an AI-dominated world? (00:42:45)Policies to avoid mass AI-layoffs (00:47:47)Who really governs Anthropic? (01:08:29)Why “pausing superintelligence” fails (01:10:52)Is American AI monopoly safe? (01:21:08)Explaining AGI to the world (01:28:40)Is Anton bullish or bearish on Germany? (01:31:05)Our production team includes:Video editors: Josh Alward, Dominic Armstrong, Ollie Bignell, Andrés Escobar, Jasper Luithlen, Milo McGuire, Luke Monsour, and Simon MonsourProducers: Elizabeth Cox and Nick StocktonCoordination and support: Katy Moore and Lou MoranCamera operator: Jeremy ChevillotteMusic: CORBIT

(0:00) The AI Buildout: Datacenters Bigger Than Cities (Andrew Feldman) (1:50) Reasoning, Inference, and Breaking Moore's Law (16:28) Open Source, AI Sovereignty, and the Road to AGI (40:54) The Innovation Behind Generative Video (Robin Rombach) (47:31) Martin Scorsese, Robots, and the Future of Hollywood IP Thanks to our partners for making this possible! AppLovin Ads - AppLovin's AI advertising platform reaches over a billion daily active users across mobile games. Full-screen video ads with a 35-second median watch time. Advertisers are profitably spending hundreds of thousands of dollars a day and advertiser access is still in closed beta. The window is open at https://applovin.com/ALLIN Nasdaq - Positioned at the nexus of technology and the capital markets, Nasdaq provides premier platforms and services for global capital markets and beyond with unmatched technology, insights and markets expertise. https://www.nasdaq.com/convergence-economy Follow Andrew: https://x.com/andrewdfeldman Follow Robin: https://x.com/robrombach Follow the besties: https://x.com/chamath https://x.com/Jason https://x.com/DavidSacks https://x.com/friedberg Follow on X: https://x.com/theallinpod Follow on Instagram: https://www.instagram.com/theallinpod Follow on TikTok: https://www.tiktok.com/@allin Follow on LinkedIn: https://www.linkedin.com/company/allinpod Intro Music Credit: https://rb.gy/tppkzl https://x.com/yung_spielburg