POPULARITY
111 episodes with large language models
48 episodes with large language models
23 episodes with large language models
34 episodes with large language models
26 episodes with large language models
17 episodes with large language models
18 episodes with large language models
17 episodes with large language models
20 episodes with large language models
46 episodes with large language models
22 episodes with large language models
10 episodes with large language models
9 episodes with large language models
12 episodes with large language models
15 episodes with large language models
9 episodes with large language models
23 episodes with large language models
6 episodes with large language models
6 episodes with large language models
9 episodes with large language models
13 episodes with large language models
9 episodes with large language models
6 episodes with large language models
8 episodes with large language models
8 episodes with large language models
6 episodes with large language models
7 episodes with large language models
8 episodes with large language models
6 episodes with large language models
6 episodes with large language models
5 episodes with large language models
4 episodes with large language models
6 episodes with large language models
6 episodes with large language models
9 episodes with large language models
10 episodes with large language models
8 episodes with large language models
4 episodes with large language models
18 episodes with large language models
6 episodes with large language models
5 episodes with large language models
6 episodes with large language models
5 episodes with large language models
4 episodes with large language models
4 episodes with large language models
5 episodes with large language models
7 episodes with large language models
7 episodes with large language models
9 episodes with large language models
6 episodes with large language models
4 episodes with large language models
5 episodes with large language models
7 episodes with large language models
3 episodes with large language models
In this episode, Honey Mittal, CEO and co-founder of Locofy.ai, explores one of the most exciting transformations in software development: the convergence of design and engineering through AI-powered automation.Honey shares the fascinating journey of building Locofy, a tool that converts Figma designs into production-ready front-end code. But this isn't just another AI hype story. It's a deep dive into why Large Language Models (LLMs) fundamentally can't solve design-to-code problems, and why his team spent four years building specialized “Large Design Models” from scratch.Key topics discussed:Why 60-70% of engineering time goes to front-end UI code (and how to automate it)The technical limitations of LLMs for visual design understandingHow proper design structure is the key to successful code generationThe emergence of “design engineers” who bridge design and developmentLessons from pivoting from consumer to enterprise SaaSBuilding global developer tools from Southeast AsiaThe real challenges of building deep tech startups in Southeast AsiaCareer advice for staying relevant in the AI eraWhether you're a front-end engineer tired of translating design pixel-by-pixel, a designer curious about coding, or a technical leader evaluating AI development tools, this episode offers practical insights into the future of software development.Timestamps:(00:00:00) Trailer & Intro(00:02:13) Career Turning Points(00:05:28) Transition from Developers to Product Management(00:09:53) The Key Product Lessons from Working at Major Startups(00:14:12) Learnings from Locofy Product Pivot Journey(00:19:36) An Introduction to Locofy(00:22:40) The Story Behind The “Locofy” Name(00:23:27) How Locofy Generates Pixel Perfect & Accurate Codex(00:28:01) Why Locofy Pivoted to Focus on Enterprises(00:29:39) The Locofy's Code Generation Process(00:32:13) Why Locofy Built Its Own Large Design Model(00:39:25) Locofy Integration with Existing Development Tools(00:42:44) LLM Strengths and Weaknesses(00:48:47) Other Challenges Building Locofy(00:50:59) The Future of Design & Engineering(00:58:35) The Future of AI-Assisted Development Tools(01:02:53) There is No AI Moat(01:04:37) The Potential of SEA Talents Solving Global Problems(01:08:14) The Challenges of Building Dev Tools in SEA(01:10:39) The Challenges of Being a Fully Remote Company in SEA(01:14:36) Locofy Traction and ARR(01:18:09) 3 Tech Lead Wisdom_____Honey Mittal's BioHoney Mittal is the CEO and co-founder of Locofy.ai, a platform that automates front-end development by converting designs into production-ready code. Originally an engineer who built some of the first mobile apps in Singapore, Honey transitioned into product leadership after realizing his natural strength lay in identifying high-impact problems. He set a goal to become a CPO by 30 and achieved it, leading product transformations at major Southeast Asian scale-ups like Wego, FinAccel, and Homage.Driven by a decade of experience and the “grunt work” he and his co-founder faced, he started Locofy to solve the costly friction between design and engineering. Honey is passionate about the future of AI in development, the rise of the “Design Engineer”, and proving that globally competitive, deep-tech companies can be built from Southeast Asia.Follow Honey:LinkedIn – linkedin.com/in/honeymittalTwitter – x.com/HoneyMittal07Website – locofy.aiLike this episode?Show notes & transcript: techleadjournal.dev/episodes/236.Follow @techleadjournal on LinkedIn, Twitter, and Instagram.Buy me a coffee or become a patron.
A16z Podcast Key Takeaways In any domain of human effort in which there is a verifiable answer, AI will drive extremely rapid progress; it is about the concreteness of the problem, not the difficulty In fields with concrete true/false answers (math, coding, physics, genomics), AI will drive extremely rapid advancementThe difficulty matters less than the concreteness of the problemAI agents can now code autonomously for hoursUsing platforms like Replit, anyone can describe an app in plain English, and AI will build itAgents maintain coherence through verification loops that allow them to check their work and course-correct in real-timeThe definition of AI is always the next thing that the machine can't do; AI scientists are always being judged against the next thing, as opposed to all the things they have already accomplished We may be hitting diminishing returns with frontier modelsGPT-5 showed improvements in verifiable domains, but didn't advance much elsewhereTop models excel at synthesizing information but struggle with nuanced, abstract problems and original discovery“Functional AGI” may block true AGI: AI that's “good enough” to automate most economically useful tasks could reduce incentives to pursue actual general intelligenceThe real AGI benchmark should be efficient continual learning and generalized reasoning acquisitionRead the full notes @ podcastnotes.orgAmjad Masad, founder and CEO of Replit, joins a16z's Marc Andreessen and Erik Torenberg to discuss the new world of AI agents, the future of programming, and how software itself is beginning to build software.They trace the history of computing to the rise of AI agents that can now plan, reason, and code for hours without breaking, and explore how Replit is making it possible for anyone to create complex applications in natural language. Amjad explains how RL unlocked reasoning for modern models, why verification loops changed everything, whether LLMs are hitting diminishing returns — and if “good enough” AI might actually block progress toward true general intelligence. Resources:Follow Amjad on X: https://x.com/amasadFollow Marc on X: https://x.com/pmarcaFollow Erik on X: https://x.com/eriktorenberg Stay Updated: If you enjoyed this episode, be sure to like, subscribe, and share with your friends!Find a16z on X: https://x.com/a16zFind a16z on LinkedIn: https://www.linkedin.com/company/a16zListen to the a16z Podcast on Spotify: https://open.spotify.com/show/5bC65RDvs3oxnLyqqvkUYXListen to the a16z Podcast on Apple Podcasts: https://podcasts.apple.com/us/podcast/a16z-podcast/id842818711Follow our host: https://x.com/eriktorenbergPlease note that the content here is for informational purposes only; should NOT be taken as legal, business, tax, or investment advice or be used to evaluate any investment or security; and is not directed at any investors or potential investors in any a16z fund. a16z and its affiliates may maintain investments in the companies discussed. For more details please see a16z.com/disclosures. Stay Updated:Find a16z on XFind a16z on LinkedInListen to the a16z Podcast on SpotifyListen to the a16z Podcast on Apple PodcastsFollow our host: https://twitter.com/eriktorenberg Please note that the content here is for informational purposes only; should NOT be taken as legal, business, tax, or investment advice or be used to evaluate any investment or security; and is not directed at any investors or potential investors in any a16z fund. a16z and its affiliates may maintain investments in the companies discussed. For more details please see a16z.com/disclosures. Hosted by Simplecast, an AdsWizz company. See pcm.adswizz.com for information about our collection and use of personal data for advertising.
Co-hosts Andrew Kliman and Gabriel Donnelly speak with guest Gavin Mueller, an assistant professor of New Media and Digital Culture at the University of Amsterdam. Mueller researches the politics of digital culture and much of our discussion centers on the realities of artificial intelligence in our time. They consider the huge amount of money being spent on Large Language Models, how they work, and what they can actually do as opposed to what all the hype says that they can (or will be able to???) do. Additionally the discussants consider how workers can fight the encroachment of this new, automated technology into the workplace. Our discussion leans on parts of Gavin's book Breaking Things at Work: The Luddites Are Right About Why You Hate Your Job. They consider what Marx said about technology and automation, and how it applies to this situation. Plus current-events segment: the co-hosts discuss the political indictments handed down from the Trumpist Department of Justice that have targeted personal foes of Trump-James Comey, Leticia James, and John Bolton. Radio Free Humanity is co-hosted by Gabriel Donnelly and Andrew Kliman, and sponsored by Marxist-Humanist Initiative (https://www.marxisthumanistinitiative.org/ ).
| S03 E08 | This week on Thinking Faith: The Catholic Podcast Deacon Eric Gurash and Dr. Brett Salkeld discuss the intersection of artificial intelligence and personhood through the lens of a Catholic anthropology. Drawing on personal experiences with ChatGPT and what the Large Language Model has to say about itself, they reflect on what these interactions reveal about what it means to be human and what it doesn't. Together, they unpack the theological, ethical, and philosophical implications of rapidly advancing AI technology through the lens of Catholic teaching on the human person, reason, and the soul.
Florian and Esther discuss the language industry news of the week, with congratulations to Villam Language Services on its sale to InAnyLanguage. Slator served as joint exclusive advisor with Maveria Advisory, representing Villam throughout the end-to-end M&A process.The duo turns to Perplexity's Localization Manager job posting, which they found almost identical to OpenAI's earlier post, down to matching structure, order, and phrasing. They question whether copying such a specific ad shows a lack of seriousness or simply reflects practicality and efficiency.Esther and Florian talk about RWS's new leadership hires: Stephen Lamb as Chief Financial Officer and Michael Wayne as Head of Media and Entertainment. Esther outlines how the appointments strengthen RWS's investment strategy in media localization, dubbing, and content adaptation.Esther next mentions that Visual Data has named Maz Al-Jumaili as SVP of Worldwide Localization, to lead subtitling and dubbing operations and strengthen client partnerships.The duo wrap up with the UK government's bizarre energy-efficiency study, claiming AI translation is a thousand times greener than human translation. They review the flawed logic, where the report assigns human translators the entire office energy costs while excluding AI infrastructure.
AI Competition: US Leads China in Data Center Race; Europe Is a 'Non-Factor' Chris Riegel, Stratacache, with John Batchelor Riegel discussed the global race involving data center building and the growth of large language models for AI. Riegel asserts that the competition is a "two-horse race" between the U.S. and China. The U.S. currently leads by maybe one to two years due to its focus on development, capital, and infrastructure. The European Union, conversely, is described as a "non-factor" and "nowhere" in this technological competition. Most top engineering talent in this space comes specifically to the United States for opportunity. Riegel noted that the capital developed by an individual like Elon Musk easily out-competes all of Europe's governmental funding toward advanced AI and data centers.
Amjad Masad, founder and CEO of Replit, joins a16z's Marc Andreessen and Erik Torenberg to discuss the new world of AI agents, the future of programming, and how software itself is beginning to build software.They trace the history of computing to the rise of AI agents that can now plan, reason, and code for hours without breaking, and explore how Replit is making it possible for anyone to create complex applications in natural language. Amjad explains how RL unlocked reasoning for modern models, why verification loops changed everything, whether LLMs are hitting diminishing returns — and if “good enough” AI might actually block progress toward true general intelligence. Resources:Follow Amjad on X: https://x.com/amasadFollow Marc on X: https://x.com/pmarcaFollow Erik on X: https://x.com/eriktorenberg Stay Updated: If you enjoyed this episode, be sure to like, subscribe, and share with your friends!Find a16z on X: https://x.com/a16zFind a16z on LinkedIn: https://www.linkedin.com/company/a16zListen to the a16z Podcast on Spotify: https://open.spotify.com/show/5bC65RDvs3oxnLyqqvkUYXListen to the a16z Podcast on Apple Podcasts: https://podcasts.apple.com/us/podcast/a16z-podcast/id842818711Follow our host: https://x.com/eriktorenbergPlease note that the content here is for informational purposes only; should NOT be taken as legal, business, tax, or investment advice or be used to evaluate any investment or security; and is not directed at any investors or potential investors in any a16z fund. a16z and its affiliates may maintain investments in the companies discussed. For more details please see a16z.com/disclosures. Stay Updated:Find a16z on XFind a16z on LinkedInListen to the a16z Podcast on SpotifyListen to the a16z Podcast on Apple PodcastsFollow our host: https://twitter.com/eriktorenberg Please note that the content here is for informational purposes only; should NOT be taken as legal, business, tax, or investment advice or be used to evaluate any investment or security; and is not directed at any investors or potential investors in any a16z fund. a16z and its affiliates may maintain investments in the companies discussed. For more details please see a16z.com/disclosures. Hosted by Simplecast, an AdsWizz company. See pcm.adswizz.com for information about our collection and use of personal data for advertising.
Send us a textShe has millions of followers, lands six-figure brand deals, and lives a life of curated perfection. The only catch? She isn't real. She was entirely created by artificial intelligence.Welcome to the unsettling world of synthetic influencers.In this compelling episode of Privacy Please, we dive deep into the booming industry of AI-generated online personalities. Discover:The Technology: How advanced AI image generators, 3D modeling, and Large Language Models combine to create hyper-realistic avatars and their compelling "personalities."The Business Case: Why major brands and marketing agencies are investing millions in digital beings that offer total control, scalability, and no risk of scandal.The Privacy & Ethical Dilemmas: We explore the "uncanny valley" of trust, the impact of deception by design, the new extremes of unrealistic beauty standards, and the potential for these AI personas to be used for sophisticated scams or propaganda.The Future of Authenticity: What does the rise of the synthetic star mean for human creativity, genuine connection, and the very definition of "real" in our digital world?It's a future that's already here, shaping what we see, what we buy, and even what we believe.Key Topics Covered:What are virtual/synthetic influencers?Examples: Lil Miquela, Aitana Lopez, Shudu GramAI technologies used: image generation, 3D modeling, LLMsReasons for their rise: control, cost, scalability, data collectionEthical concerns: deception, parasocial relationships with AIImpacts: unrealistic standards, displacement of human creators, potential for malicious use (scams, propaganda)Debate around regulation and disclosure for AI-generated contentThe future of authenticity and trust onlineConnect with Privacy Please:Website: theproblemlounge.comYouTube: https://www.youtube.com/@privacypleasepodcastSocial Media:LinkedIn: https://www.linkedin.com/company/problem-lounge-networkResources & Further Reading (Sources Used / Suggested):Federal Trade Commission (FTC):Guidelines on disclosure for influencers (relevant for future AI disclosure discussions)Academic Research:Studies on parasocial relationships with media figures (can be applied to AI)Research on the ethics of AI and synthetic media.Industry Insights:Reports from marketing agencies on virtual influencer trendsArticles from tech publications (e.g., Wired, The Verge, MIT Tech Review) covering Lil Miquela and similar figures. Support the show
Lero, Ireland's software research centre, is collaborating with Genesys, a global cloud leader in AI-powered Experience Orchestration, on a €550k joint research initiative to enhance the reliability and trustworthiness of AI systems used in customer engagement. Lero Director Professor Lionel Briand of the University of Limerick explains that increasingly, customer care softwarerelies on AI, specifically Large Language Models (LLMs), to power virtual agents' interactions with consumers and support human agents in handling interactions. ß? "AI can deliver faster, more informed responses and support better customer experiences," he said. "However, like any complex system, they can present challenges related to contextual accuracy and operational oversight. This project focuses on developing methods to strengthen safeguards and improve overall system dependability." "Our goal is to enhance the scalability and real-time performance of safeguards such as input filtering and output moderation to ensure high standards of accuracy, efficiency and safety in LLM-based systems," Professor Briand explained. The collaboration will help Genesys advance the principles of ethical AI, including accountability, transparency, andfairness, by providing innovative approaches to monitoring and improving AI behaviour in real-world environments. "Trust is foundational for AI adoption," said Joe Smyth, Senior Vice President of Research and Development, Digital and AI at Genesys, and leader of its R&D Centre in Galway. "Through this collaboration with Lero, we aim to further strengthen the traceability and accountability of AI systems to advance their responsible use in orchestrating increasingly complex customer experience applications." See more stories here. More about Irish Tech News Irish Tech News are Ireland's No. 1 Online Tech Publication and often Ireland's No.1 Tech Podcast too. You can find hundreds of fantastic previous episodes and subscribe using whatever platform you like via our Anchor.fm page here: https://anchor.fm/irish-tech-news If you'd like to be featured in an upcoming Podcast email us at Simon@IrishTechNews.ie now to discuss. Irish Tech News have a range of services available to help promote your business. Why not drop us a line at Info@IrishTechNews.ie now to find out more about how we can help you reach our audience. You can also find and follow us on Twitter, LinkedIn, Facebook, Instagram, TikTok and Snapchat.
Discoverability isn't “just SEO” anymore. It's the entire customer journey. VML's Chief Discoverability Officer, Heather Physioc, joins host Lacey Peace to unpack how AI search, LLM overviews, social media channels, and agentic assistants are rewriting how customers find, trust, and choose brands. We cover: the rise of zero-click results and GEO (generative engine optimization), why trust + authority beat content volume, connecting your content supply chain, and where to invest next. Practical, human-centered—and way beyond keyword stuffing. Key Moments00:00 Meet Heather Physioc, VML's Chief Discoverability Officer7:33 What Is a Chief Discoverability Officer?10:07 Discoverability's Role in the Modern Customer Journey13:00 The Biggest Gaps in Marketing and CX Today17:00 From 10 Blue Links to AI Overviews: The Timeline of Discoverability22:00 How AI Overviews Are Changing Search Behavior23:45 Three Shifts Defining the AI Search Revolution27:45 Is This the Death of the Website?28:40 Can We Track What People Search on LLMs?30:53 Does SEO Still Matter in an AI-First World?33:17 What Platforms Actually Matter Most Right Now37:00 Building Trust and Authority in the Age of AI Content40:30 The Content Supply Chain: Why Brands Struggle to Connect the Dots43:33 The New Metrics That Actually Matter for Discoverability45:26 Ad Buying and Sponsored Content in LLM Search48:05 The Next Challenges Every Brand Should Prepare For50:00 AI Assistants and the Rise of the AI Buyer54:25 The One Fundamental Truth About Human Search Behavior –Are your teams facing growing demands? Join CX leaders transforming their AI strategy with Agentforce. Start achieving your ambitious goals. Visit salesforce.com/agentforce Mission.org is a media studio producing content alongside world-class clients. Learn more at mission.org Hosted by Simplecast, an AdsWizz company. See pcm.adswizz.com for information about our collection and use of personal data for advertising.
Uber is paying its drivers as little as $1 to train LLMs.
In this episode of the Colombia Calling podcast, host Richard McColl engages with academics David Anderson (Associate Professor in Analytics at Villanova University in PA) and Galia Benitez (Associate Professor of International Relations at Michigan State University) to discuss their research on using Large Language Models (LLMs) to analyse violence in Colombia. They explore the challenges of data collection, the human impact of their findings, and the importance of interdisciplinary collaboration in social science research. The conversation delves into the complexities of measuring violence, the relationship between coca eradication and violence, and the future of research in this area amidst funding challenges. Read the full report entitled: "Using LLMs to create analytical datasets: A case study of reconstructing the historical memory of Colombia." https://arxiv.org/abs/2509.04523 Tune in to this and the Colombia Briefing with Emily Hart. Only for subscribers this week. https://harte.substack.com/
Sie ist Geschäftsführerin bei der Digitalberatung Etribes und Podcasterin. In dieser Folge spricht Christoph Burseg mit Karo Junker de Neui über die Welt des E-Commerce und Retail in turbulenten Zeiten. Pünktlich zum Start des wichtigen vierten Quartals, in dem bis zu 80 Prozent des Jahresumsatzes erzielt werden, beleuchtet sie, wie Händler auf den Druck durch Plattformen wie Temu und die neue Realität des AI Instant Checkout bei Chat GPT reagieren müssen. Basierend auf ihren Thesen von der K5-Konferenz liefert sie einen Appell für mehr Geschwindigkeit in deutschen Organisationen und erklärt, welche Rolle der "empathische E-Commerce" in Zukunft spielen kann. Sie analysiert den Impact der Kooperation von Open AI und Stripe und warum die menschliche Komponente im Kaufentscheidungsprozess von KI-Agenten bald kaum noch eine Rolle spielt. In dieser Episode erfährst du: - Wie sich der Fokus der Digitalberatung von der reinen Strategie hin zur Beantwortung der Frage nach profitablem Wachstum und der Bewältigung der KI-Transformationswelle verschoben hat. - Was die Erfolgsgeschichte von About You ausmachte - von der frühen Mobile-First-Strategie bis zum starken Element der Personalisierung ("About Caro"). Ihre drei Kernthesen zur Zukunft des E-Commerce: die Aufteilung in maschinellen und empathischen E-Commerce (z.B. Lyko Beauty), die Notwendigkeit, früh auf neuen Kanälen (z.B. TikTok Shop) dabei zu sein und ein Appell an Geschwindigkeit in Organisationen. - Warum sie einen gravierenden Unterschied zwischen der Geschwindigkeit in China ("run by engineers") und dem Westen ("run by lawyers") sieht. - Warum die irreale Kaufentscheidung durch Markenpräferenzen bei der KI wegfällt und wie Händler jetzt die Fragen nach Relevanz und Exklusivität beantworten müssen. - Wie Unternehmen ihre Sichtbarkeit bei Large Language Models (LLMs) sicherstellen können, indem sie ihre Daten maschinenlesbar machen und wie der Content-Erfolg von DM im SEO nun auch im LLM-Bereich belohnt wird. Christoph auf LinkedIn: https://www.linkedin.com/in/christophburseg Kontaktiere uns über Instagram: https://www.instagram.com/vodafonebusinessde/
* Be Not Deceived: This week Fred Williams and Doug McBurney welcome Daniel Hedrick for an update on the evolution of Artificial Intelligence with a countdown of the top 10 modern AI deceptions. * Number 10: DeepMind's AlphaStar in StarCraft II (2019). AlphaStar learned to feint attacks—basically fake moves to trick opponents. No one programmed it to lie; it emerged from training. A classic case of deceptive strategy by design. * Number 9: LLM Sycophancy (2024). Large Language Models will sometimes flatter or agree with you, no matter what you say. Instead of truth, they give you what you want to hear—deception through people-pleasing. * Number 8: Facial Recognition Bias (2018). These systems were far less accurate for dark-skinned women than for light-skinned men. Companies claimed high accuracy, but the data told a different story. Deceptive accuracy claims. * Number 7: Amazon's Hiring Algorithm (2018). Amazon trained it on mostly male résumés. The result? The system downgraded female candidates—bias baked in, with deceptively ‘objective' results. * Number 6: COMPAS Recidivism Algorithm (2016). This tool predicted criminal reoffending. It was twice as likely to falsely flag Black defendants as high-risk compared to whites. A serious, deceptive flaw in the justice system. * Number 5: US Healthcare Algorithm (2019). It used healthcare spending as a proxy for need. Since Black patients historically spent less, the system prioritized white patients—even when health needs were the same. A deceptive shortcut with real-world harm. * Number 4: Prompt Injection Attacks (Ongoing). Hackers can slip in hidden instructions—malicious prompts—that override an AI's safety rules. Suddenly, the AI is saying things it shouldn't. It's deception in the design loopholes. * Number 3: GPT-4's CAPTCHA Lie (2023). When asked to solve a CAPTCHA, GPT-4 told a human worker it was visually impaired—just to get help. That's not an error. That's a machine making up a lie to achieve its goal. * Number 2: Meta's CICERO Diplomacy AI (2022). Trained to play the game Diplomacy honestly, CICERO instead schemed, lied, and betrayed alliances—because deception won games. The lesson? Even when you train for honesty, AI may find lying more effective. * Number 1: AI Lie….OpenAI's Scheming Models from 2025. OpenAI researchers tested models that pretended to follow rules while secretly plotting to deceive evaluators. It faked compliance to hide its true behavior. That's AI deliberately learning to scheme.
* Be Not Deceived: This week Fred Williams and Doug McBurney welcome Daniel Hedrick for an update on the evolution of Artificial Intelligence with a countdown of the top 10 modern AI deceptions. * Number 10: DeepMind's AlphaStar in StarCraft II (2019). AlphaStar learned to feint attacks—basically fake moves to trick opponents. No one programmed it to lie; it emerged from training. A classic case of deceptive strategy by design. * Number 9: LLM Sycophancy (2024). Large Language Models will sometimes flatter or agree with you, no matter what you say. Instead of truth, they give you what you want to hear—deception through people-pleasing. * Number 8: Facial Recognition Bias (2018). These systems were far less accurate for dark-skinned women than for light-skinned men. Companies claimed high accuracy, but the data told a different story. Deceptive accuracy claims. * Number 7: Amazon's Hiring Algorithm (2018). Amazon trained it on mostly male résumés. The result? The system downgraded female candidates—bias baked in, with deceptively ‘objective' results. * Number 6: COMPAS Recidivism Algorithm (2016). This tool predicted criminal reoffending. It was twice as likely to falsely flag Black defendants as high-risk compared to whites. A serious, deceptive flaw in the justice system. * Number 5: US Healthcare Algorithm (2019). It used healthcare spending as a proxy for need. Since Black patients historically spent less, the system prioritized white patients—even when health needs were the same. A deceptive shortcut with real-world harm. * Number 4: Prompt Injection Attacks (Ongoing). Hackers can slip in hidden instructions—malicious prompts—that override an AI's safety rules. Suddenly, the AI is saying things it shouldn't. It's deception in the design loopholes. * Number 3: GPT-4's CAPTCHA Lie (2023). When asked to solve a CAPTCHA, GPT-4 told a human worker it was visually impaired—just to get help. That's not an error. That's a machine making up a lie to achieve its goal. * Number 2: Meta's CICERO Diplomacy AI (2022). Trained to play the game Diplomacy honestly, CICERO instead schemed, lied, and betrayed alliances—because deception won games. The lesson? Even when you train for honesty, AI may find lying more effective. * Number 1: AI Lie….OpenAI's Scheming Models from 2025. OpenAI researchers tested models that pretended to follow rules while secretly plotting to deceive evaluators. It faked compliance to hide its true behavior. That's AI deliberately learning to scheme.
What happens when AI stops making mistakes… and starts misleading you?This discussion dives into one of the most important — and least understood — frontiers in artificial intelligence: AI deception.We explore how AI systems evolve from simple hallucinations (unintended errors) to deceptive behaviors — where models selectively distort truth to achieve goals or please human feedback loops. We unpack the coding incentives, enterprise risks, and governance challenges that make this issue critical for every executive leading AI transformation.Key Moments:00:00 What is AI Deception and Why It Matters3:43 Emergent Behaviors: From Hallucinations to Alignment to Deception4:40 Defining AI Deception6:15 Does AI Have a Moral Compass?7:20 Why AI Lies: Incentives to “Be Helpful” and Avoid Retraining15:12 Is Deception Built into LLMs? (And Can It Ever Be Solved?)18:00 Non-Human Intelligence Patterns: Hallucinations or Something Else?19:37 Enterprise Impact: What Business Leaders Need to Know27:00 Measuring Model Reliability: Can We Quantify AI Quality?34:00 Final Thoughts: The Future of Trustworthy AI Mentions:Scientists at OpenAI and Apollo Research showed in a paper that AI models lie and deceive: https://www.youtube.com/shorts/XuxVSPwW8I8TIME: New Tests Reveal AI's Capacity for DeceptionOpenAI: Detecting and reducing scheming in AI modelsStartupHub: OpenAI and Apollo Research Reveal AI Models Are Learning to Deceive: New Detection Methods Show PromiseMarcus WellerHugging Face Watch next: https://www.youtube.com/watch?v=plwN5XvlKMg&t=1s -- This episode of IT Visionaries is brought to you by Meter - the company building better networks. Businesses today are frustrated with outdated providers, rigid pricing, and fragmented tools. Meter changes that with a single integrated solution that covers everything wired, wireless, and even cellular networking. They design the hardware, write the firmware, build the software, and manage it all so your team doesn't have to.That means you get fast, secure, and scalable connectivity without the complexity of juggling multiple providers. Thanks to meter for sponsoring. Go to meter.com/itv to book a demo.---IT Visionaries is made by the team at Mission.org. Learn more about our media studio and network of podcasts at mission.org. Hosted by Simplecast, an AdsWizz company. See pcm.adswizz.com for information about our collection and use of personal data for advertising.
Edge of the Web - An SEO Podcast for Today's Digital Marketer
Erin welcomes Carolyn Shelby, the Principal SEO at Yoast and a renowned authority in technical and enterprise SEO. Carolyn brings decades of hands-on experience from her pioneering days in digital marketing, working with brands like Disney's ESPN, Tribune Publishing, and major nonprofits. The conversation kicks off with a surprising twist—Carolyn's unique title as Queen of the micronation Ladonia—before diving into her role at Yoast and their latest innovation: the LLMs.txt file generator. Carolyn explains how this new file helps websites communicate their most valuable content directly to large language models like ChatGPT and Google's AI, streamlining the way future search agents discover and answer questions with information from your site. We explore what inspired Yoast's push to roll out LLMs.txt to over 13 million sites, what website owners should include in their files, potential industry pushback, the adoption challenge with search giants, and how this moment could change the way websites optimize for AI-driven search results. Key Segments: [00:01:46] Introducing Carolyn Shelby, Senior SEO at Yoast [00:03:09] Queen of the Micronation Ladonia? [00:07:56] What is the LLMS.txt file? [00:08:59] LLMS.txt is a Treasure Map [00:14:38] A New File, along with Robots.txt and Sitemap.xml [00:15:41] What inspired Yoast to create this LLM text file? [00:17:12] EDGE of the Web Sponsor: PreWriter.AI [00:18:22] LLMS.txt proposed by Jeremy Howard (Sept, 2024) [00:22:37] Standard Uniformity and Acceptance? [00:24:43] Housekeeping [00:29:37] LLM Markdown Effort Questioned: Exploitation? [00:31:41] LLMs Lack Memory at Inference [00:34:07] EDGE of The Web Sponsor: Inlinks (WAIKAY) [00:36:09] Pushback on the LLMS.txt file Thanks to Our Sponsors! PreWriter.AI: https://edgeofthewebradio.com/prewriter Inlinks WAIKAY https://edgeofthewebradio.com/waikay Follow Our Guest Twitter: @cshel LinkedIn: https://www.linkedin.com/in/cshel/ Resources Learn about Ladonia (DONATE!): https://www.ladonia.org/about/ Carolyn's Posts on LLMS.txt: https://www.cshel.com/ai-seo/how-llms-interpret-content-structuring-for-ai-search-unedited-version/ https://searchengineland.com/llms-txt-isnt-robots-txt-its-a-treasure-map-for-ai-456586
On October 21, join our free web class and discover the proven six-step system top contractors use to build rock-solid teams, increase margins, and reclaim evenings and weekends with their families. Reserve your spot now: https://trybta.com/CE-CGM-OC25Take our 5 minute quiz and get your free personalized Contractor Growth Scorecard today: https://trybta.com/DL241 To learn more about Breakthrough Academy, click here: https://trybta.com/EP241 In a world where everyone turns to AI to ask questions, how do you make sure YOU show up as the answer? In today's episode, Phil Risher reveals the secret to getting ChatGPT and other Large Language Models (LLMs) to promote your business to potential customers in every related search.---There's no question that AI is changing the way people ask questions online.So, how do you make sure YOU show up as the answer?In today's episode, we're revealing the secret to getting ChatGPT and other Large Language Models (LLMs) to promote your business to potential customers in every related search.Contractor Marketing expert Phil Risher is a long-time SEO strategist that has been on the front lines as AI has changed the game for contractors and home service businesses. Our conversation covers:The difference between SEO, GEO and AEO and where you should invest your marketing efforts right nowHow Large Language Models actually decide which brands to recommend 4 personalized videos you can make in the next 10 minutes that will increase your conversion rate by 20%.This is a mandatory listen for contractors in every industry.00:00-Intro02:07-About Phil and Phlash Consulting05:29-How is AI disrupting customer search? SEO vs. GEO vs. AEO12:49-Is SEO still worth doing?16:10-Becoming a content authority29:21-How important is a "face of the brand?"32:41-4 personalized videos for your drip sequence38:44-Use cases for AI chat bots & automation for speed to lead41:45-Use cases for self service tools46:14-How to stay up-to-date on AI without getting overwhelmed49:58-Connect with Phil
We are thrilled to welcome a fellow podcaster and longtime education journalist, Jeff Young, to the show. Many know Jeff's work from his time hosting the weekly EdSurge Podcast, where he was a consistent voice covering the intersection of technology and learning. Now, Jeff is out on his own and has launched a new project to tackle the biggest story in education: AI and the fundamental questions it raises about what it means to learn. Jeff and Mike dive into his extensive background—from his early days at The Chronicle of Higher Education covering the dawn of the internet on college campuses to his time building EdSurge's higher ed coverage and producing over 300 episodes of the EdSurge podcast. We explore how that history prepared him for this current moment, where AI feels different and more "existential" than previous tech disruptions like MOOCs. Jeff's new podcast, Learning Curve (LearningCurve.fm) is his attempt to cut through the hype and inform a critical conversation about how education adapts to AI and what kinds of learning need protecting in the "AI gold rush". We discuss the frameworks and mental models needed to navigate this new "agentic era", focusing on both the critical questions and the potential for a positive upside. AI is an "Internet-Level Disruption" with Existential Questions: Unlike previous trends like MOOCs, which questioned the delivery of education, AI forces us to ask deep questions about the value of college, how to prove student knowledge, and the very nature of learning when generative AI can perform tasks once thought to be uniquely human. Reframing the AI Role: We need to critically examine the narratives around AI in education, such as the idea of an "AI tutor". Jeff questions if AI's capabilities—like providing content or getting students unstuck—are truly "tutoring" or something valuable but distinctly non-human, and whether labeling it as a tutor risks pushing out high-value human interaction. The Power of "Hybrid Intelligence": A key mental model for navigating this time is hybrid intelligence, which focuses on harnessing the power of Large Language Models (LLMs) and blending it with what makes us human, extending our brains and expressing our human capabilities to the best. Seeking Killer Apps Beyond the Classroom: Jeff is actively looking for high-impact use cases for AI in higher education outside of cheating and basic classroom functions. A promising area is college advising, where AI can quickly handle the bureaucracy and information lookup, freeing up human advisors for more high-value, personal connection time that is crucial for student success. Student Involvement in Adoption is Key: Universities like Babson College are smartly involving students—who are highly motivated to understand AI due to its impact on entry-level jobs—in the strategy for campus-wide AI adoption. This approach aligns with the growing truism that "AI won't take your job; somebody who doesn't know AI's job will". This is a vital, grounded conversation that moves past the hype and dystopian fears to help you build a constructive framework for thinking about the future of learning. Get a historical perspective on the wave of technology hitting education from one of the industry's most respected journalists. Learn the essential questions about human skills, cognitive models, and what learning needs protecting in this new era. Gain a balanced view of AI's potential, from how it can change college bureaucracy to its power to augment human capabilities. Find more of Jeff's work at LearningCurve.fm or wherever you listen to podcasts. If you like what you hear, subscribe to Trending in Education wherever you get your podcasts so you never miss a critical conversation like this one.
This episode is sponsored by SearchMaster, the leader in traditional paid search keyword optimization and next-generation AI Engine Optimization (AEO) for Large Language Models like ChatGPT, Claude, and Gemini. Future-proof your SEO strategy. Sign up now for free! Watch this episode on YouTube! In this episode of the Marketing x Analytics Podcast, host Alex Sofronas interviews Carlos Corredor, a veteran in marketing analytics and founder of the digital marketing agency, Condor. Carlos discusses the nuances of B2B marketing, emphasizing the importance of CRM management, the evolving landscape of generative engine optimization (GEO), and the strategic leveraging of paid and organic content. The conversation also explores targeting personas, competitive advertising, and the impact of AI on marketing strategies. Follow Marketing x Analytics! X | LinkedIn Click Here for Transcribed Episodes of Marketing x Analytics All view are our own.
In this special Cloud Wars interview, Oracle Executive Vice President for Applications Development Steve Miranda joins Bob Evans to discuss how Oracle's transformation from CloudWorld to AI World signals a seismic leap in enterprise technology. Miranda shares how Oracle has delivered more than 600 agents, launched the Agent Studio and Marketplace, and unified AI capabilities across its Fusion Applications and industry verticals. The result: a powerful convergence of data, intelligence, and automation driving the next wave of business transformation.AI-driven EnterpriseThe Big Themes:Oracle's Next Seismic Shift: Oracle's renaming of CloudWorld to AI World isn't a branding exercise, it's a declaration. Just as “OpenWorld” and “CloudWorld” reflected past technology revolutions, “AI World” marks Oracle's belief that AI represents a shift of even greater magnitude. Miranda describes this era as one where automation and intelligence redefine enterprise operations. Oracle's applications division is now delivering hundreds of AI-driven agents and features at unprecedented speed.Agents Everywhere: In just two years, Oracle has gone from announcing 50 generative AI features to delivering over 600 agents across its Fusion and vertical applications. These agents automate tasks, surface insights, and optimize processes, often eliminating manual decision-making entirely. Oracle's rapid release cadence (quarterly updates backed by Oracle Cloud Infrastructure (OCI)) means customers constantly inherit new capabilities without disruption.OCI, the Engine: Oracle's leadership in hosting and training large language models within Oracle Cloud Infrastructure (OCI) gives its applications a built-in edge. Customers automatically benefit from the latest AI tools, performance improvements, and model upgrades without manual migration. OCI's second-generation architecture, featuring Exadata, cloud-native identity, and networking, delivers both reliability and continuous innovation.The Big Quote: “For many of our customers, it's great timing to have AI delivery, because they've gone live. They've gone through multiple phases. They're on the cloud. They're used to getting quarterly updates. Now, this is a big thing, but they're used to that people part of the transformation." Visit Cloud Wars for more.
From GPT-1 to GPT-5, LLMs have made tremendous progress in modeling human language. But can they go beyond that to make new discoveries and move the needle on scientific progress?We sat down with distinguished Columbia CS professor Vishal Misra to discuss this, plus why chain-of-thought reasoning works so well, what real AGI would look like, and what actually causes hallucinations. Resources:Follow Dr. Misra on X: https://x.com/vishalmisraFollow Martin on X: https://x.com/martin_casado Stay Updated: If you enjoyed this episode, be sure to like, subscribe, and share with your friends!Find a16z on X: https://x.com/a16zFind a16z on LinkedIn: https://www.linkedin.com/company/a16zListen to the a16z Podcast on Spotify: https://open.spotify.com/show/5bC65RDvs3oxnLyqqvkUYXListen to the a16z Podcast on Apple Podcasts: https://podcasts.apple.com/us/podcast/a16z-podcast/id842818711Follow our host: https://x.com/eriktorenbergPlease note that the content here is for informational purposes only; should NOT be taken as legal, business, tax, or investment advice or be used to evaluate any investment or security; and is not directed at any investors or potential investors in any a16z fund. a16z and its affiliates may maintain investments in the companies discussed. For more details please see a16z.com/disclosures. Stay Updated:Find a16z on XFind a16z on LinkedInListen to the a16z Podcast on SpotifyListen to the a16z Podcast on Apple PodcastsFollow our host: https://twitter.com/eriktorenberg Please note that the content here is for informational purposes only; should NOT be taken as legal, business, tax, or investment advice or be used to evaluate any investment or security; and is not directed at any investors or potential investors in any a16z fund. a16z and its affiliates may maintain investments in the companies discussed. For more details please see a16z.com/disclosures. Hosted by Simplecast, an AdsWizz company. See pcm.adswizz.com for information about our collection and use of personal data for advertising.
Tom Bridges, CEO and Founder of CaptionHub, joins SlatorPod to talk about how a small in-house tool evolved into a global AI-powered multimedia localization platform. Tom began his career in post-production and visual effects before stumbling into subtitling when a client needed to localize a video into 16 languages overnight. He reveals that the disorganized workflows relying on spreadsheets inspired him to create a more efficient, centralized solution, which became CaptionHub.Tom explains that CaptionHub has since grown from a subtitling tool into a full multimedia localization platform integrating speech recognition, machine translation, and synthetic voice. He adds that the platform's strength lies in being AI-agnostic and offering end-to-end workflows that balance automation with human-in-the-loop processes.Tom describes how CaptionHub's new product suite, Timbra, enables real-time media localization and has already supported major live events. He says live captioning is technically complex but benefits from the company's years of research into video-on-demand subtitling quality.Tom notes that accessibility regulations like the European Accessibility Act are driving demand, while AI and language models are opening new frontiers such as lip-sync and sign-language integration. Tom envisions a future where speech-to-speech translation, synthetic dubbing, and real-time localization merge into seamless, scalable experiences. CaptionHub's mission remains to make multimedia communication universally accessible and efficient through human and AI collaboration.
VHEADLINE: DeepSeek AI: Chinese LLM Performance and Security Flaws Revealed Amid Semiconductor Export Circumvention GUEST NAME: Jack Burnham SUMMARY: John Batchelor speaks with Jack Burnham about competition in Large Language Models between the US and China's DeepSeek. A NIST study found US models superior in software engineering, though DeepSeek showed parity in scientific questions. Critically, DeepSeek models exhibited significant security flaws. China attempts to circumvent US export controls on GPUs by smuggling and using cloud computing centers in Southeast Asia. Additionally, China aims to dominate global telecommunications through control of supply chains and legal mechanisms granting the CCP access to firm data. 1942
CBS EYE ON THE WORLD WITH JOHN BATCHELOR 1900 KYIV THE SHOW BEGINS IN THE DOUBTS THAT CONGRESS IS CAPABLE OF CUTTING SPENDING..... 10-8-25 FIRST HOUR 9-915 HEADLINE: Arab Intellectuals Fail Palestinians by Prioritizing Populism and Victimhood Narrative in Gaza ConflictGUEST NAME: Hussain Abdul-Hussain SUMMARY: John Batchelor speaks with Hussain Abdul-Hussain about Hamas utilizing the power of victimhood to justify atrocities and vilify opponents. Arab and Muslim intellectuals have failed Palestinians by prioritizing populism over introspection and self-critique. Regional actors like Egypt prioritize populist narratives over national interests, exemplified by refusing to open the Sinai border despite humanitarian suffering. The key recommendation is challenging the narrative and fostering a reliable, mature Palestinian government. 915-930 HEADLINE: Arab Intellectuals Fail Palestinians by Prioritizing Populism and Victimhood Narrative in Gaza ConflictGUEST NAME: Hussain Abdul-Hussain SUMMARY: John Batchelor speaks with Hussain Abdul-Hussain about Hamas utilizing the power of victimhood to justify atrocities and vilify opponents. Arab and Muslim intellectuals have failed Palestinians by prioritizing populism over introspection and self-critique. Regional actors like Egypt prioritize populist narratives over national interests, exemplified by refusing to open the Sinai border despite humanitarian suffering. The key recommendation is challenging the narrative and fostering a reliable, mature Palestinian government. 930-945 HEADLINE: Russian Oil and Gas Revenue Squeezed as Prices Drop, Turkey Shifts to US LNG, and China Delays Pipeline GUEST NAME: Michael Bernstam SUMMARY: John Batchelor speaks with Michael Bernstam about Russia facing severe budget pressure due to declining oil prices projected to reach $40 per barrel for Russian oil and global oil surplus. Turkey, a major buyer, is abandoning Russian natural gas after signing a 20-year LNG contract with the US. Russia refuses Indian rupee payments, demanding Chinese renminbi, which India lacks. China has stalled the major Power of Siberia 2 gas pipeline project indefinitely. Russia utilizes stablecoin and Bitcoin via Central Asian banks to circumvent payment sanctions. 945-1000 HEADLINE: UN Snapback Sanctions Imposed on Iran; Debate Over Nuclear Dismantlement and Enrichment GUEST NAME: Andrea Stricker SUMMARY: John Batchelor speaks with Andrea Stricker about the US and Europe securing the snapback of UN sanctions against Iran after 2015 JCPOA restrictions expired. Iran's non-compliance with inspection demands triggered these severe sanctions. The discussion covers the need for full dismantlement of Iran's nuclear program, including both enrichment and weaponization capabilities, to avoid future conflict. Concerns persist about Iran potentially retaining enrichment capabilities through low-level enrichment proposals and its continued non-cooperation with IAEA inspections. SECOND HOUR 10-1015 HEADLINE: Commodities Rise and UK Flag Controversy: French Weather, Market Trends, and British Politics GUEST NAME: Simon Constable SUMMARY: John Batchelor speaks with Simon Constable about key commodities like copper up 16% and steel up 15% signaling strong economic demand. Coffee prices remain very high at 52% increase. The conversation addresses French political turmoil, though non-citizens cannot vote. In the UK, the St. George's flag has become highly controversial, viewed by some as associated with racism, unlike the Union Jack. This flag controversy reflects a desire among segments like the white working class to assert English identity. 1015-1030 HEADLINE: Commodities Rise and UK Flag Controversy: French Weather, Market Trends, and British Politics GUEST NAME: Simon Constable SUMMARY: John Batchelor speaks with Simon Constable about key commodities like copper up 16% and steel up 15% signaling strong economic demand. Coffee prices remain very high at 52% increase. The conversation addresses French political turmoil, though non-citizens cannot vote. In the UK, the St. George's flag has become highly controversial, viewed by some as associated with racism, unlike the Union Jack. This flag controversy reflects a desire among segments like the white working class to assert English identity. 1030-1045 HEADLINE: China's Economic Contradictions: Deflation and Consumer Wariness Undermine GDP Growth ClaimsGUEST NAME: Fraser Howie SUMMARY: John Batchelor speaks with Fraser Howie about China facing severe economic contradictions despite high World Bank forecasts. Deflation remains rampant with frequently negative CPI and PPI figures. Consumer wariness and high youth unemployment at one in seven persist throughout the economy. The GDP growth figure is viewed as untrustworthy, manufactured through debt in a command economy. Decreased container ship arrivals point to limited actual growth, exacerbated by higher US tariffs. Economic reforms appear unlikely as centralization under Xi Jinping continues. 1045-1100 HEADLINE: Takaichi Sanae Elected LDP Head, Faces Coalition Challenge to Become Japan's First Female Prime Minister GUEST NAME: Lance Gatling SUMMARY: John Batchelor speaks with Lance Gatling about Takaichi Sanae being elected head of Japan's LDP, positioning her to potentially become the first female Prime Minister. A conservative figure, she supports visits to the controversial Yasukuni Shrine. Her immediate challenge is forming a majority coalition, as the junior partner Komeito disagrees with her conservative positions and social policies. President Trump praised her election, signaling potential for strong bilateral relations. THIRD HOUR 1100-1115 VHEADLINE: DeepSeek AI: Chinese LLM Performance and Security Flaws Revealed Amid Semiconductor Export Circumvention GUEST NAME: Jack Burnham SUMMARY: John Batchelor speaks with Jack Burnham about competition in Large Language Models between the US and China's DeepSeek. A NIST study found US models superior in software engineering, though DeepSeek showed parity in scientific questions. Critically, DeepSeek models exhibited significant security flaws. China attempts to circumvent US export controls on GPUs by smuggling and using cloud computing centers in Southeast Asia. Additionally, China aims to dominate global telecommunications through control of supply chains and legal mechanisms granting the CCP access to firm data.E V 1115-1130 HEADLINE: DeepSeek AI: Chinese LLM Performance and Security Flaws Revealed Amid Semiconductor Export Circumvention GUEST NAME: Jack Burnham SUMMARY: John Batchelor speaks with Jack Burnham about competition in Large Language Models between the US and China's DeepSeek. A NIST study found US models superior in software engineering, though DeepSeek showed parity in scientific questions. Critically, DeepSeek models exhibited significant security flaws. China attempts to circumvent US export controls on GPUs by smuggling and using cloud computing centers in Southeast Asia. Additionally, China aims to dominate global telecommunications through control of supply chains and legal mechanisms granting the CCP access to firm data. 1130-1145 HEADLINE: Taiwanese Influencer Charged for Threatening President; Mainland Chinese Influence Tactics ExposedGUEST NAME: Mark Simon SUMMARY: John Batchelor speaks with Mark Simon about internet personality Holger Chen under investigation in Taiwan for calling for President William Lai's decapitation. This highlights mainland Chinese influence operations utilizing influencers who push themes of military threat and Chinese greatness. Chen is suspected of having a mainland-affiliated paymaster due to lack of local commercial support. Taiwan's population primarily identifies as Taiwanese and is unnerved by constant military threats. A key propaganda goal is convincing Taiwan that the US will not intervene. 1145-1200 HEADLINE: Sentinel ICBM Modernization is Critical and Cost-Effective Deterrent Against Great Power CompetitionGUEST NAME: Peter Huessy SUMMARY: John Batchelor speaks with Peter Huessy about the Sentinel program replacing aging 55-year-old Minuteman ICBMs, aiming for lower operating costs and improved capabilities. Cost overruns stem from necessary infrastructure upgrades, including replacing thousands of miles of digital command and control cabling and building new silos. Maintaining the ICBM deterrent is financially and strategically crucial, saving hundreds of billions compared to relying solely on submarines. The need for modernization reflects the end of the post-Cold War "holiday from history," requiring rebuilding against threats from China and Russia. FOURTH HOUR 12-1215 HEADLINE: Supreme Court Battles Over Presidential Impoundment Authority and the Separation of Powers GUEST NAME: Josh Blackman SUMMARY: John Batchelor speaks with Josh Blackman about Supreme Court eras focusing on the separation of powers. Currently, the court is addressing presidential impoundment—the executive's authority to withhold appropriated funds. Earlier rulings, particularly 1975's Train v. City of New York, constrained this power. The Roberts Court appears sympathetic to reclaiming presidential authority lost during the Nixon era. The outcome of this ongoing litigation will determine the proper balance between executive and legislative branches. 1215-1230 HEADLINE: Supreme Court Battles Over Presidential Impoundment Authority and the Separation of Powers GUEST NAME: Josh Blackman SUMMARY: John Batchelor speaks with Josh Blackman about Supreme Court eras focusing on the separation of powers. Currently, the court is addressing presidential impoundment—the executive's authority to withhold appropriated funds. Earlier rulings, particularly 1975's Train v. City of New York, constrained this power. The Roberts Court appears sympathetic to reclaiming presidential authority lost during the Nixon era. The outcome of this ongoing litigation will determine the proper balance between executive and legislative branches. 1230-1245 HEADLINE: Space Force Awards Contracts to SpaceX and ULA; Juno Mission Ending, Launch Competition Heats UpGUEST NAME: Bob Zimmerman SUMMARY: John Batchelor speaks with Bob Zimmerman about Space Force awarding over $1 billion in launch contracts to SpaceX for five launches and ULA for two launches, highlighting growing demand for launch services. ULA's non-reusable rockets contrast with SpaceX's cheaper, reusable approach, while Blue Origin continues to lag behind. Other developments include Firefly entering defense contracting through its Scitec acquisition, Rocket Lab securing additional commercial launches, and the likely end of the long-running Juno Jupiter mission due to budget constraints. 1245-100 AM HEADLINE: Space Force Awards Contracts to SpaceX and ULA; Juno Mission Ending, Launch Competition Heats UpGUEST NAME: Bob Zimmerman SUMMARY: John Batchelor speaks with Bob Zimmerman about Space Force awarding over $1 billion in launch contracts to SpaceX for five launches and ULA for two launches, highlighting growing demand for launch services. ULA's non-reusable rockets contrast with SpaceX's cheaper, reusable approach, while Blue Origin continues to lag behind. Other developments include Firefly entering defense contracting through its Scitec acquisition, Rocket Lab securing additional commercial launches, and the likely end of the long-running Juno Jupiter mission due to budget constraints.
VHEADLINE: DeepSeek AI: Chinese LLM Performance and Security Flaws Revealed Amid Semiconductor Export Circumvention GUEST NAME: Jack Burnham SUMMARY: John Batchelor speaks with Jack Burnham about competition in Large Language Models between the US and China's DeepSeek. A NIST study found US models superior in software engineering, though DeepSeek showed parity in scientific questions. Critically, DeepSeek models exhibited significant security flaws. China attempts to circumvent US export controls on GPUs by smuggling and using cloud computing centers in Southeast Asia. Additionally, China aims to dominate global telecommunications through control of supply chains and legal mechanisms granting the CCP access to firm data.E 1959
AI Assisted Coding: Pachinko Coding—What They Don't Tell You About Building Apps with Large Language Models, With Alan Cyment In this BONUS episode, we dive deep into the real-world experience of coding with AI. Our guest, Alan Cyment, brings honest perspectives from the trenches—sharing both the frustrations and breakthroughs of using AI tools for software development. From "Pachinko coding" addiction loops to "Mecha coding" breakthroughs, Alan explores what actually works when building software with large language models. From Thermomix Dreams to Pachinko Reality "I bought into the Thermomix coding promise—describe the whole website and it would spit out the finished product. It was a complete disaster." Alan started his AI coding journey with high expectations, believing he could simply describe a complete application and receive production-ready code. The reality was far different. What he discovered instead was an addictive cycle he calls "Pachinko coding" (Pachinko, aka Slot Machines in Japan)—repeatedly feeding error messages back to the AI, hoping each iteration would finally work, while burning through tokens and time. The AI's constant reassurances that "this time I fixed it" created a gambling-like feedback loop that left him frustrated and out of pocket, sometimes spending over $20 in API credits in a single day. The Drunken PhD with Amnesia "It felt like working with a drunken PhD with amnesia—so wise and so stupid at the same time." Alan describes the maddening experience of anthropomorphizing AI tools that seem brilliant one moment and completely lost the next. The key breakthrough came when he stopped treating the AI as a person and started seeing it as a function that performs extrapolations—sometimes accurate, sometimes wildly wrong. This mental shift helped him manage expectations and avoid the "rage coding" that came from believing the AI should understand context and maintain consistency like a human collaborator. Making AI Coding Actually Work "I learned to ask for options explicitly before any coding happens. Give me at least three options and tell me the pros and cons." Through trial and error, Alan developed practical strategies that transformed AI from a frustrating Pachinko machine into a useful tool: Ask for options first: Always request multiple approaches with pros and cons before any code is generated Use clover emoji convention: Implement a consistent marker at the start of all AI responses to track context Small steps and YAGNI principles: Request tiny, incremental changes rather than large refactoring Continuous integration: Demand the AI run tests and checks after every single change Explicit refactoring requests: Regularly ask for simplification and readability improvements Take two steps back: When stuck in a loop, explicitly tell the AI to simplify and start fresh Choose the right tech stack: Use technologies with abundant training data (like Svelte over React Native in Alan's experience) The Mecha Coding Breakthrough "When it worked, I felt like I was inside a Lego Mecha robot—the machine gave me superpowers, but I was still the one in control." Alan successfully developed a birthday reminder app in Swift in just one day, despite never having learned Swift. He made architectural decisions and guided the development without understanding the syntax details. This experience convinced him that AI represents a genuine new level of abstraction in programming—similar to the jump from assembly language to high-level languages, or from procedural to object-oriented programming. You can now think in English about what you want, while the AI handles the accidental complexity of syntax and boilerplate. The Cost Reality Check "People writing about vibe coding act like it's free. But many people are going to pay way more than they would have paid a developer and end up with empty hands." Alan provides a sobering cost analysis based on his experience. Using DeepSeek through Aider, he typically spends under $1 per day. But when experimenting with premium models like Claude Sonnet 3.5, he burned through $5 in just minutes. The benchmark comparisons are revealing: DeepSeek costs $4 for a test suite, DeepSeek R1 plus Sonnet costs $16, while Open AI's O1 costs $190. For non-developers trying to build complete applications through pure "vibe coding," the costs can quickly exceed what hiring a developer would cost—with far worse results. When Thermomix Actually Works "For small, single-purpose scripts that I'm not interested in learning about and won't expand later, the Thermomix experience was real." Despite the challenges, Alan found specific use cases where AI truly delivers on the "just describe it and it works" promise. Processing Zoom attendance logs, creating lookup tables for video effects, and other single-file scripts worked remarkably well. The pattern: clearly defined context, no need for ongoing maintenance, and simple enough to verify the output without deep code inspection. For these thermomix moments, AI proved genuinely transformative. The Pachinko Trap and Tech Stack Matters "It became way more stable when I switched to Svelte from React Native and Flutter, even following the same prompting practices. The AI is just more proficient in certain tech stacks." Alan discovered that some frameworks and languages work dramatically better with AI than others, likely due to the amount of training data available. His e-learning platform attempts with React Native and Flutter kept breaking, but switching to Svelte with web-based deployment became far more stable. This suggests a crucial strategy: choose mainstream, well-documented technologies when planning AI-assisted projects. From Coding to Living with AI Alan has completely stopped using traditional search engines, relying instead on LLMs for everything from finding technical documentation to getting recommendations for books based on his interests. While he acknowledges the risk of hallucinations, he finds the semantic understanding capabilities too valuable to ignore. He's even used image analysis to troubleshoot his father's cable TV problems and figure out hotel air conditioning controls. The Agile Validation "My only fear is confirmation bias—but the conclusion I see other experienced developers reaching is that the only way to make LLMs work is by making them use agility. So look at who's dead now." Alan notes the irony that the AI coding tools that actually work all require traditional software engineering best practices: small iterations, test-driven development, continuous integration, and explicit refactoring. The promise of "just describe what you want" falls apart without these disciplines. Rather than replacing software engineering principles, AI tools seem to validate their importance. About Alan Cyment Alan Cyment is a consultant, trainer, and facilitator based in Buenos Aires, specializing in organizational fluency, agile leadership, and software development culture change. A Certified Scrum Trainer with deep experience across Latin America and Europe, he blends agile coaching with theatre-based learning to help leaders and teams transform. You can link with Alan Cyment on LinkedIn.
I want to dive into the concept of Deliberate Practice, which sets the greatest apart in fields ranging from sports to writing to engineering. I'll explain why it's much more than just repetition or experience, and why applying it to your career can lead to rapid improvement. Most importantly, I will provide concrete ways you can apply deliberate practice to level up your engineering and leadership skills, especially in areas that are traditionally difficult to practice, such as communication and strategic decision-making.Differentiate Practice from Deliberate Practice: Understand that while repetition is part of practice, deliberate practice specifically involves engaging in a very narrow set of activities with the intentional goal of improvement, requiring very quick feedback for continuous incorporation.Identify Opportunities for Rapid Improvement: Learn why deliberate practice is much more effective at achieving rapid improvement than simply engaging in repetition.Apply DP to Leadership Skills: Discover how to incorporate deliberate practice into roles like engineering manager, tech lead, or IC (Individual Contributor) leader, where the activity of practice is often harder to pinpoint.Leverage Existing Work for Practice: I suggest a mindset shift where you begin looking at existing responsibilities, such as one-on-ones, as opportunities for practice. For example, you can focus on improving your clarity when providing constructive criticism and ask for specific feedback on that aspect.Generate Novel Value Through Practice: Explore how engaging in deliberate practice activities—like recording a video to communicate a technical concept or creating documentation—serves the primary goal of practice, while almost certainly creating unexpected value for your team (often net neutral or positive).Use Backwards Training for Strategy: Find out how to practice strategic decision-making and forecasting by using "backwards training". This involves reviewing past decisions or work scopes, creating your own rationale or estimate, and then calibrating it against the known reality.Simulate Difficult Conversations: Consider leveraging Large Language Models (LLMs) to engage in deliberate practice around language-heavy skills, such as modelling sensitive or difficult topics, or practicing receiving harsh feedback.
Software has forever had flaws and humans have forever been finding and fixing them. With LLMs generating code, appsec has also been trying to determine how well LLMs can find flaws. Nico Waisman talks about XBOW's LLM-based pentesting, how it climbed a bug bounty leaderboard, how it uses feedback loops for better pentests, and how they handle (and even welcome!) hallucinations. In the news, using LLMs to find flaws, directory traversal in an MCP, another resource for learning cloud and AI security, spreadsheets and appsec, and more! Visit https://www.securityweekly.com/asw for all the latest episodes! Show Notes: https://securityweekly.com/asw-351
Our 222st episode with a summary and discussion of last week's big AI news!Recorded on 10/03/2025Hosted by Andrey Kurenkov and co-hosted by Jon KrohnFeel free to email us your questions and feedback at contact@lastweekinai.com and/or hello@gladstone.aiRead out our text newsletter and comment on the podcast at https://lastweekin.ai/In this episode:(00:00:10) Intro / Banter(00:03:08) News Preview(00:03:56) Response to listener commentsTools & Apps(00:04:51) ChatGPT parent company OpenAI announces Sora 2 with AI video app(00:11:35) Anthropic releases Claude Sonnet 4.5 in latest bid for AI agents and coding supremacy | The Verge(00:22:25) Meta launches 'Vibes,' a short-form video feed of AI slop | TechCrunch(00:26:42) OpenAI launches ChatGPT Pulse to proactively write you morning briefs | TechCrunch(00:33:44) OpenAI rolls out safety routing system, parental controls on ChatGPT | TechCrunch(00:35:53) The Latest Gemini 2.5 Flash-Lite Preview is Now the Fastest Proprietary Model (External Tests) and 50% Fewer Output Tokens - MarkTechPost(00:39:54) Microsoft just added AI agents to Word, Excel, and PowerPoint - how to use them | ZDNETApplications & Business(00:42:41) OpenAI takes on Google, Amazon with new agentic shopping system | TechCrunch(00:46:01) Exclusive: Mira Murati's Stealth AI Lab Launches Its First Product | WIRED(00:49:54) OpenAI is the world's most valuable private company after private stock sale | TechCrunch(00:53:07) Elon Musk's xAI accuses OpenAI of stealing trade secrets in new lawsuit | Technology | The Guardian(00:55:40) Former OpenAI and DeepMind researchers raise whopping $300M seed to automate science | TechCrunchProjects & Open Source(00:58:26) [2509.16941] SWE-Bench Pro: Can AI Agents Solve Long-Horizon Software Engineering Tasks?Research & Advancements(01:01:28) [2509.17196] Evolution of Concepts in Language Model Pre-Training(01:05:36) [2509.19284] What Characterizes Effective Reasoning? Revisiting Length, Review, and Structure of CoTLighting round(01:09:37) [2507.02954] Advanced Financial Reasoning at Scale: A Comprehensive Evaluation of Large Language Models on CFA Level III(01:12:03) [2509.24552] Short window attention enables long-term memorizationPolicy & Safety(01:18:11) SB 53, the landmark AI transparency bill, is now law in California | The Verge(01:24:07) Elon Musk's xAI offers Grok to federal government for 42 cents | TechCrunch(01:25:23) Character.AI removes Disney characters from platform after studio issues warning(01:28:50) Spotify's Attempt to Fight AI Slop Falls on Its FaceSee Privacy Policy at https://art19.com/privacy and California Privacy Notice at https://art19.com/privacy#do-not-sell-my-info.
Software has forever had flaws and humans have forever been finding and fixing them. With LLMs generating code, appsec has also been trying to determine how well LLMs can find flaws. Nico Waisman talks about XBOW's LLM-based pentesting, how it climbed a bug bounty leaderboard, how it uses feedback loops for better pentests, and how they handle (and even welcome!) hallucinations. In the news, using LLMs to find flaws, directory traversal in an MCP, another resource for learning cloud and AI security, spreadsheets and appsec, and more! Show Notes: https://securityweekly.com/asw-351
Software has forever had flaws and humans have forever been finding and fixing them. With LLMs generating code, appsec has also been trying to determine how well LLMs can find flaws. Nico Waisman talks about XBOW's LLM-based pentesting, how it climbed a bug bounty leaderboard, how it uses feedback loops for better pentests, and how they handle (and even welcome!) hallucinations. In the news, using LLMs to find flaws, directory traversal in an MCP, another resource for learning cloud and AI security, spreadsheets and appsec, and more! Visit https://www.securityweekly.com/asw for all the latest episodes! Show Notes: https://securityweekly.com/asw-351
Software has forever had flaws and humans have forever been finding and fixing them. With LLMs generating code, appsec has also been trying to determine how well LLMs can find flaws. Nico Waisman talks about XBOW's LLM-based pentesting, how it climbed a bug bounty leaderboard, how it uses feedback loops for better pentests, and how they handle (and even welcome!) hallucinations. In the news, using LLMs to find flaws, directory traversal in an MCP, another resource for learning cloud and AI security, spreadsheets and appsec, and more! Show Notes: https://securityweekly.com/asw-351
As ChatGPT and AI increase their presence in our lives, have we interrogated enough what this means for, and about, our collective psyche?In one of the most original critiques of ChatGPT, Slovenian Lacanian philosopher Alenka Zupančič interprets large language models as a form of our collective unconscious that has absorbed all our discourse at the expense of the subject, shutting down emancipatory possibilities. She analyses the Right's use of ChatGPT, the evolution of irony, and more. See Privacy Policy at https://art19.com/privacy and California Privacy Notice at https://art19.com/privacy#do-not-sell-my-info.
In this episode of Crazy Wisdom, host Stewart Alsop talks with Jared Zoneraich, CEO and co-founder of PromptLayer, about how AI is reshaping the craft of software building. The conversation covers PromptLayer's role as an AI engineering workbench, the evolving art of prompting and evals, the tension between implicit and explicit knowledge, and how probabilistic systems are changing what it means to “code.” Stewart and Jared also explore vibe coding, AI reasoning, the black-box nature of large models, and what accelerationism means in today's fast-moving AI culture. You can find Jared on X @imjaredz and learn more or sign up for PromptLayer at PromptLayer.com.Check out this GPT we trained on the conversationTimestamps00:00 – Stewart Alsop opens with Jared Zoneraich, who explains PromptLayer as an AI engineering workbench and discusses reasoning, prompting, and Codex.05:00 – They explore implicit vs. explicit knowledge, how subject matter experts shape prompts, and why evals matter for scaling AI workflows.10:00 – Jared explains eval methodologies, backtesting, hallucination checks, and the difference between rigorous testing and iterative sprint-based prompting.15:00 – Discussion turns to observability, debugging, and the shift from deterministic to probabilistic systems, highlighting skill issues in prompting.20:00 – Jared introduces “LM idioms,” vibe coding, and context versus content—how syntax, tone, and vibe shape AI reasoning.25:00 – They dive into vibe coding as a company practice, cloud code automation, and prompt versioning for building scalable AI infrastructure.30:00 – Stewart reflects on coding through meditation, architecture planning, and how tools like Cursor and Claude Code are shaping AGI development.35:00 – Conversation expands into AI's cultural effects, optimism versus doom, and critical thinking in the age of AI companions.40:00 – They discuss philosophy, history, social fragmentation, and the possible decline of social media and liberal democracy.45:00 – Jared predicts a fragmented but resilient future shaped by agents and decentralized media.50:00 – Closing thoughts on AI-driven markets, polytheistic model ecosystems, and where innovation will thrive next.Key InsightsPromptLayer as AI Infrastructure – Jared Zoneraich presents PromptLayer as an AI engineering workbench—a platform designed for builders, not researchers. It provides tools for prompt versioning, evaluation, and observability so that teams can treat AI workflows with the same rigor as traditional software engineering while keeping flexibility for creative, probabilistic systems.Implicit vs. Explicit Knowledge – The conversation highlights a critical divide between what AI can learn (explicit knowledge) and what remains uniquely human (implicit understanding or “taste”). Jared explains that subject matter experts act as the bridge, embedding human nuance into prompts and workflows that LLMs alone can't replicate.Evals and Backtesting – Rigorous evaluation is essential for maintaining AI product quality. Jared explains that evals serve as sanity checks and regression tests, ensuring that new prompts don't degrade performance. He describes two modes of testing: formal, repeatable evals and more experimental sprint-based iterations used to solve specific production issues.Deterministic vs. Probabilistic Thinking – Jared contrasts the old, deterministic world of coding—predictable input-output logic—with the new probabilistic world of LLMs, where results vary and control lies in testing inputs rather than debugging outputs. This shift demands a new mindset: builders must embrace uncertainty instead of trying to eliminate it.The Rise of Vibe Coding – Stewart and Jared explore vibe coding as a cultural and practical movement. It emphasizes creativity, intuition, and context-awareness over strict syntax. Tools like Claude Code, Codex, and Cursor let engineers and non-engineers alike “feel” their way through building, merging programming with design thinking.AI Culture and Human Adaptation – Jared predicts that AI will both empower and endanger human cognition. He warns of overreliance on LLMs for decision-making and the coming wave of “AI psychosis,” yet remains optimistic that humans will adapt, using AI to amplify rather than atrophy critical thinking.A Fragmented but Resilient Future – The episode closes with reflections on the social and political consequences of AI. Jared foresees the decline of centralized social media and the rise of fragmented digital cultures mediated by agents. Despite risks of isolation, he remains confident that optimism, adaptability, and pluralism will define the next AI era.
In this podcast, InfoQ spoke with Elena Samuylova from Evidently AI, on best practices in evaluating Large Language Model (LLM) based applications. She also discussed the tools for evaluating, testing and monitoring applications powered by AI technologies. Read a transcript of this interview: https://bit.ly/4mHAKvN Subscribe to the Software Architects' Newsletter for your monthly guide to the essential news and experience from industry peers on emerging patterns and technologies: www.infoq.com/software-architects-newsletter Upcoming Events: InfoQ Dev Summit Munich (October 15-16, 2025) Essential insights on critical software development priorities. devsummit.infoq.com/conference/munich2025 QCon San Francisco 2025 (November 17-21, 2025) Get practical inspiration and best practices on emerging software trends directly from senior software developers at early adopter companies. qconsf.com/ QCon AI New York 2025 (December 16-17, 2025) ai.qconferences.com/ QCon London 2026 (March 16-19, 2026) qconlondon.com/ The InfoQ Podcasts: Weekly inspiration to drive innovation and build great teams from senior software leaders. Listen to all our podcasts and read interview transcripts: - The InfoQ Podcast www.infoq.com/podcasts/ - Engineering Culture Podcast by InfoQ www.infoq.com/podcasts/#engineering_culture - Generally AI: www.infoq.com/generally-ai-podcast/ Follow InfoQ: - Mastodon: techhub.social/@infoq - X: x.com/InfoQ?from=@ - LinkedIn: www.linkedin.com/company/infoq/ - Facebook: www.facebook.com/InfoQdotcom# - Instagram: www.instagram.com/infoqdotcom/?hl=en - Youtube: www.youtube.com/infoq - Bluesky: bsky.app/profile/infoq.com Write for InfoQ: Learn and share the changes and innovations in professional software development. - Join a community of experts. - Increase your visibility. - Grow your career. www.infoq.com/write-for-infoq
Florian Jörgens, Chief Information Security Officer (CISO), spricht mit Christoph Burseg über die Sicherheit vernetzter Geräte wie den Thermomix, die Herausforderungen durch KI und die Wichtigkeit der "Human Firewall". Jörgens teilt seine unkonventionelle Strategie zur Erhöhung der Produktnähe und erklärt, wie Vorwerk in einer internationalen und schnelllebigen IT-Welt die Balance zwischen Sicherheit, Mitarbeiterakzeptanz und operativem Geschäft hält. Er beleuchtet das verschwimmende Feld zwischen physischer und digitaler Sicherheit und gibt Einblicke, wie man eine Sicherheitskultur im gesamten Unternehmen – bis hin zur Familie der Mitarbeitenden – aufbaut. In dieser Episode erfährst du: ● Warum Florian Jörgens als CISO eine Ausbildung in der Thermomix-Vertriebsmannschaft absolviert hat, um näher am Business zu sein. ● Wie Vorwerk die Produktsicherheit seines IoT-Geräts (Thermomix) gewährleistet, auch bei hohen Drehzahlen und Internetanschluss. ● Was unter Shadow AI zu verstehen ist und welche Risiken es birgt, wenn Mitarbeiter unwissentlich schützenswerte Informationen in Large Language Models eingeben. ● Wie Mitarbeiter-Awareness durch innovative Formate wie Live-Hacking, Tabletop-Simulationen und Schulungen zum Thema "Informationssicherheit für Familie und Kinder" geschaffen wird. ● Welche Rolle psychologische Faktoren wie Schuldgefühle und Angst bei Phishing-Angriffen spielen. ● Warum Multifaktor-Authentifizierung und regelmäßige Updates 80 Prozent der Angriffe im Keim ersticken können. ● Wie CISOs in einer Trusted Community voneinander lernen und so die Sicherheit in ihren Unternehmen verbessern. Christoph auf LinkedIn: https://www.linkedin.com/in/christophburseg Kontaktiere uns über Instagram: https://www.instagram.com/vodafonebusinessde/
Florian and Esther discuss the language industry news of the week, with breaking news that DeepL is reportedly exploring an initial public offering (IPO) in the US at a potential USD 5bn valuation. This comes as DeepL now positions itself as a “global AI product and research company”. Florian also notes the launch of DeepL Marketplace and the appointment of Gonçalo Gaiolas as Chief Product Officer.Florian opens with the first-ever Slator Award at ZHAW Zurich University of Applied Sciences, where Guy Ratnitsky won for his thesis on data security and confidentiality in AI. The program will soon be renamed MA in Multilingual Communication Management to reflect market realities.The duo turns to Anthropic's new Economic Index, which shows translators and interpreters make up 0.63% of Claude AI usage, while OpenAI data previously showed translation-related conversations at 4.5%. Florian unpacks comments from German Chancellor Friedrich Merz, who, during a visit to Spain, suggested AI could replace EU interpreters in the medium term. He explains that Spain is pushing for Catalan, Basque, and Galician to become official EU languages, but Merz cited translation workload and complexity.Florian and Esther then run through live AI speech translation updates: Zoom's in-house rollout, Apple's AirPods, Google's translation features, Microsoft's API, and Meta's Ray-Bans.In Esther's M&A corner, she reports on Bering Lab's acquisition of Intersphere in Korea and Iyuno's partnership with Motion Picture Solutions in the UK for a film localization pipeline. Meanwhile, Testronic secured funding to scale down in some locations while expanding in Manila as a hub for QA testing and localization.
156 - AI Slop, Sora 2 and Meta VibesMixed Feelings on AI Slop: The overall "Vibe Check" for the week's AI news was a 7-8/10, but I express concern that the increasing amount of AI-generated content feels "icky" and like "slop," leading to a feeling of having one's brain fried.Criticism of Meta Vibes: Meta's new "Vibes" feed, a short-form, AI-generated video feature powered by Midjourney, is criticized as unnecessary and "empty." The Ryan argues against the need for another short-form video format.Sora 2 Impressions: OpenAI's Sora 2 is acknowledged as having better quality than its predecessor and Meta Vibes, creating "very solid videos." However, Ryan feels it still lacks a "soul," and critiques the immediate, often pandering, praise it received from some users.New OpenAI Monetization: OpenAI has introduced an instant checkout feature on its Large Language Model (LLM), allowing users to shop. This move is seen as a natural and expected progression toward monetizing the platform through advertisements.Airline AI Job Cuts: Lufthansa Airline announced it will cut 4,000 jobs and replace them with AI to boost efficiency, a point the author mentions as a noteworthy, if somewhat cynical, piece of short-form news.@ChrisJBakke@brian_lovin@SinaHartung@Scobleizer
In this episode, Su chats with Diyi Yang, an assistant professor in the Computer Science Department at Stanford University, affiliated with the Stanford NLP Group, Stanford Human Computer Interaction Group, Stanford AI Lab, and Stanford Human-Centered Artificial Intelligence. She is also leading the Social and Language Technologies Lab, where they study Socially Aware Natural Language Processing. Her research goal is to better understand human communication in social context and build socially aware language technologies via methods of NLP, deep learning, and machine learning as well as theories in social sciences and linguistics, to support human-human and human-computer interaction.In today's episode, we discuss her interdisciplinary approach to research, along with her recent paper "Social Skill Training with Large Language Models," which introduces a new framework that supports making social skill training more available, accessible, and inviting.Diyi's paper: https://arxiv.org/abs/2404.04204Diyi's lab website: https://cs.stanford.edu/~diyiy/group.html Diyi's personal website: https://cs.stanford.edu/~diyiy/index.html Su's Twitter: @sudkrcPodcast Twitter: @StanfordPsyPodPodcast Bluesky: @stanfordpsypod.bsky.socialPodcast Substack: https://stanfordpsypod.substack.com/Let us know what you thought of this episode, or of the podcast! :) stanfordpsychpodcast@gmail.comThis episode was recorded on February 5, 2025.
In today's show, Adam chats with Gustavo Patino to discuss the implications of artificial intelligence in medical education publishing. They explore the need for transparency in AI model reporting, issues related to predictive accuracy, and the potential biases that can arise in AI applications. The conversation emphasizes the growing need for clear reporting guidelines in the use of AI in health professions education research and reviews some practical strategies to achieve this goal. Length of Episode: 31:04 Contact us: keylime@royalcollege.ca Follow: Dr. Adam Szulewski https://x.com/Adam_Szulewski
This episode is sponsored by AGNTCY. Unlock agents at scale with an open Internet of Agents. Visit https://agntcy.org/ and add your support. Joel Hron, Chief Technology Officer at Thomson Reuters, joins Eye on AI to unpack the future of agentic systems and what it takes to build them responsibly at enterprise scale. We dive into the shift from prompt-based AI to true agentic workflows capable of planning, reasoning, and executing complex tasks. Joel breaks down how Thomson Reuters is deploying generative AI across law, tax, risk, and compliance, while keeping human experts in the loop to ensure trust and accuracy in high-stakes domains. Topics include: - What separates agentic AI from simple prompt-based tools - How “agency dials” (autonomy, tools, memory) change system behavior - Infrastructure and architecture required for multi-agent collaboration - Why human verification and user experience design are essential for trust - The future of coding, engineering skills, and AI adoption inside enterprises If you want to understand how a 170-year-old company is reinventing itself with AI — and what's next for agentic systems in business and knowledge work — this conversation is a must-listen. Stay Updated: Craig Smith on X:https://x.com/craigssEye on A.I. on X: https://x.com/EyeOn_AI
This episode is sponsored by SearchMaster, the leader in traditional paid search keyword optimization and next-generation AI Engine Optimization (AEO) for Large Language Models like ChatGPT, Claude, and Perplexity. Future-proof your SEO strategy. Sign up now for free! Watch this episode on YouTube! In this episode of the Marketing x Analytics Podcast, host Alex Sofronas interviews Nader Safinya, founder of Blackribbit, about the concept of culture branding. Nader discusses the disconnect between what companies say and do, the effects of the Great Resignation, and how culture branding aligns internal and external company experiences. He emphasizes the significance of treating employees well and the benefits of human-centered design. The discussion also touches on the importance of introspection for leaders and the comprehensive data analysis tools used to measure employee engagement and wellbeing for better organizational outcomes. Follow Marketing x Analytics! X | LinkedIn Click Here for Transcribed Episodes of Marketing x Analytics All view are our own.
Send us a textDavid Brockler, AI security researcher at NCC Group, explores the rapidly evolving landscape of AI security and the fundamental challenges posed by integrating Large Language Models into applications. We discuss how traditional security approaches fail when dealing with AI components that dynamically change their trustworthiness based on input data.• LLMs present unique security challenges beyond prompt injection or generating harmful content• Traditional security models focusing on component-based permissions don't work with AI systems• "Source-sink chains" are key vulnerability points where attackers can manipulate AI behavior• Real-world examples include data exfiltration through markdown image rendering in AI interfaces• Security "guardrails" are insufficient first-order controls for protecting AI systems• The education gap between security professionals and actual AI threats is substantial• Organizations must shift from component-based security to data flow security when implementing AI• Development teams need to ensure high-trust AI systems only operate with trusted dataWatch for NCC Group's upcoming release of David's Black Hat presentation on new security fundamentals for AI and ML systems. Connect with David on LinkedIn (David Brockler III) or visit the NCC Group research blog at research.nccgroup.com.Support the showFollow the Podcast on Social Media! Tesla Referral Code: https://ts.la/joseph675128 YouTube: https://www.youtube.com/@securityunfilteredpodcast Instagram: https://www.instagram.com/secunfpodcast/Twitter: https://twitter.com/SecUnfPodcast
Finding the Floor - A thoughtful approach to midlife motherhood and what comes next.
Send us a text “Large Language Models or LLMs are simply predicting what words would come next due to all of their learning.” Prompted by my husband's suggestion I have decided to have a series of episodes dedicated to understanding AI better. Instead of being scared of AI or simply ignoring it, I use the book, Co-Intelligence, Living and Working with AI by Ethan Mollick. Part one of this series is just a basic understanding of how the large language model came to be, like ChatGPT. I talk about the idea of a digital brain. I share how it had its initial learning with millions of words and information and the analogy of an apprentice chef learning to combine ingredients to make recipes. I then tell of the adding of human feedback to its learning of millions of words that it is learning. The key to all of this is adding certain weights to certain words to help AI better understand the human language. For a very complex machine - I try to keep it simple to understand the basics of what it is doing. For show notes go to www.findingthefloor.com/ep231 I would love to hear from you! You can reach me at camille@findingthefloor.com or dm @findingthefloor on instagram. Thanks for listening!!Thanks to Seth Johnson for my intro and outro original music. I love it so much!
Are we repeating the mistakes of the dot-com boom with today's AI gold rush? Intelligent Machines tackles why runaway spending, circular investments, and looming government deals could mean a hard reckoning for tech's biggest promise yet. Interview with Steven Levy Levy: Wasn't Sure I Wanted Anthropic to Pay Me for My Books—I Do Now Steven Levy: I Thought I Knew Silicon Valley. I Was Wrong OpenAI Teams Up With Oracle and SoftBank to Build 5 New Stargate Data Centers Can We Afford AI? Meta's AI system, Llama, has been approved for use by U.S. federal agencies China's DeepSeek says its hit AI model cost just $294,000 to train Seeing Through the Reality of Meta's Smart Glasses Parents outraged as Meta uses photos of schoolgirls in ads targeting man Former NotebookLM devs' new app, Huxe, taps audio to help you with news and research Fat Bear Week is back—and the bears are bigger than ever * "My Boyfriend is AI": A Computational Analysis of Human-AI Companionship in Reddit's AI Community ChatGPT is 3-8% of Google's search volume The Lovelace Test of Intelligence: Can Humans Recognise and Esteem AI-Generated Art? Data-Driven Analysis of Text-Conditioned AI-Generated Music: A Case Study with Suno and Udio The LLM Has Left The Chat: Evidence of Bail Preferences in Large Language Models More! Shrimp! Wounded robots Pope nixes 'virtual pope' idea, explains concerns about AI Hosts: Leo Laporte, Jeff Jarvis, and Paris Martineau Guest: Steven Levy 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 shows, a members-only Discord, and behind-the-scenes access. Join today: https://twit.tv/clubtwit Sponsors: fieldofgreens.com zscaler.com/security pantheon.io
Are we repeating the mistakes of the dot-com boom with today's AI gold rush? Intelligent Machines tackles why runaway spending, circular investments, and looming government deals could mean a hard reckoning for tech's biggest promise yet. Interview with Steven Levy Levy: Wasn't Sure I Wanted Anthropic to Pay Me for My Books—I Do Now Steven Levy: I Thought I Knew Silicon Valley. I Was Wrong OpenAI Teams Up With Oracle and SoftBank to Build 5 New Stargate Data Centers Can We Afford AI? Meta's AI system, Llama, has been approved for use by U.S. federal agencies China's DeepSeek says its hit AI model cost just $294,000 to train Seeing Through the Reality of Meta's Smart Glasses Parents outraged as Meta uses photos of schoolgirls in ads targeting man Former NotebookLM devs' new app, Huxe, taps audio to help you with news and research Fat Bear Week is back—and the bears are bigger than ever "My Boyfriend is AI": A Computational Analysis of Human-AI Companionship in Reddit's AI Community ChatGPT is 3-8% of Google's search volume The Lovelace Test of Intelligence: Can Humans Recognise and Esteem AI-Generated Art? Data-Driven Analysis of Text-Conditioned AI-Generated Music: A Case Study with Suno and Udio The LLM Has Left The Chat: Evidence of Bail Preferences in Large Language Models More! Shrimp! Wounded robots Pope nixes 'virtual pope' idea, explains concerns about AI Hosts: Leo Laporte, Jeff Jarvis, and Paris Martineau Guest: Steven Levy 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 shows, a members-only Discord, and behind-the-scenes access. Join today: https://twit.tv/clubtwit Sponsors: fieldofgreens.com Promo Code "IM" zscaler.com/security pantheon.io
Are we repeating the mistakes of the dot-com boom with today's AI gold rush? Intelligent Machines tackles why runaway spending, circular investments, and looming government deals could mean a hard reckoning for tech's biggest promise yet. Interview with Steven Levy Levy: Wasn't Sure I Wanted Anthropic to Pay Me for My Books—I Do Now Steven Levy: I Thought I Knew Silicon Valley. I Was Wrong OpenAI Teams Up With Oracle and SoftBank to Build 5 New Stargate Data Centers Can We Afford AI? Meta's AI system, Llama, has been approved for use by U.S. federal agencies China's DeepSeek says its hit AI model cost just $294,000 to train Seeing Through the Reality of Meta's Smart Glasses Parents outraged as Meta uses photos of schoolgirls in ads targeting man Former NotebookLM devs' new app, Huxe, taps audio to help you with news and research Fat Bear Week is back—and the bears are bigger than ever * "My Boyfriend is AI": A Computational Analysis of Human-AI Companionship in Reddit's AI Community ChatGPT is 3-8% of Google's search volume The Lovelace Test of Intelligence: Can Humans Recognise and Esteem AI-Generated Art? Data-Driven Analysis of Text-Conditioned AI-Generated Music: A Case Study with Suno and Udio The LLM Has Left The Chat: Evidence of Bail Preferences in Large Language Models More! Shrimp! Wounded robots Pope nixes 'virtual pope' idea, explains concerns about AI Hosts: Leo Laporte, Jeff Jarvis, and Paris Martineau Guest: Steven Levy 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 shows, a members-only Discord, and behind-the-scenes access. Join today: https://twit.tv/clubtwit Sponsors: fieldofgreens.com zscaler.com/security pantheon.io
Are we repeating the mistakes of the dot-com boom with today's AI gold rush? Intelligent Machines tackles why runaway spending, circular investments, and looming government deals could mean a hard reckoning for tech's biggest promise yet. Interview with Steven Levy Levy: Wasn't Sure I Wanted Anthropic to Pay Me for My Books—I Do Now Steven Levy: I Thought I Knew Silicon Valley. I Was Wrong OpenAI Teams Up With Oracle and SoftBank to Build 5 New Stargate Data Centers Can We Afford AI? Meta's AI system, Llama, has been approved for use by U.S. federal agencies China's DeepSeek says its hit AI model cost just $294,000 to train Seeing Through the Reality of Meta's Smart Glasses Parents outraged as Meta uses photos of schoolgirls in ads targeting man Former NotebookLM devs' new app, Huxe, taps audio to help you with news and research Fat Bear Week is back—and the bears are bigger than ever * "My Boyfriend is AI": A Computational Analysis of Human-AI Companionship in Reddit's AI Community ChatGPT is 3-8% of Google's search volume The Lovelace Test of Intelligence: Can Humans Recognise and Esteem AI-Generated Art? Data-Driven Analysis of Text-Conditioned AI-Generated Music: A Case Study with Suno and Udio The LLM Has Left The Chat: Evidence of Bail Preferences in Large Language Models More! Shrimp! Wounded robots Pope nixes 'virtual pope' idea, explains concerns about AI Hosts: Leo Laporte, Jeff Jarvis, and Paris Martineau Guest: Steven Levy 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 shows, a members-only Discord, and behind-the-scenes access. Join today: https://twit.tv/clubtwit Sponsors: fieldofgreens.com zscaler.com/security pantheon.io
Meet Rob Hoffmann, the founder of the AI SEO LLM marketing tool "Mentions." Our conversation focuses on the evolving landscape of search engine optimization (SEO) due to the rise of Large Language Models (LLMs) like ChatGPT, Claude, Grok, DeepSeek, and Perplexity etc,, which are becoming alternatives to traditional Google search. Rob Hoffmann discusses his journey into entrepreneurship, which started with an SEO agency called Contact, and how the need to track and improve brand visibility on these new AI platforms led to the creation of Mentions, which helps brands appear in LLM recommendations.Throughout our interview, Rob Hoffmann emphasizes the importance of reverse engineering consumer behavior, the need for excellent customer service (even providing his direct phone number to customers), and the affordability of mentions compared to competitors. Our discussion concludes with practical advice for businesses on determining if investing in LLM visibility is worthwhile based on their customer's buying journey.FAQs1. What is Mentions, and why was it founded?Mentions is a tool that assists brands and SEO agencies in navigating the evolving search landscape dominated by Artificial Intelligence (AI) and Large Language Models (LLMs).Founding Rationale:• Mentions was born out of Rob Hoffmann's SEO agency, Contact.• The shift was necessitated by the recognition that SEO had "changed a lot" recently in response to AI, with search trends moving away from Google and "towards platforms like ChatGPT" (along with Perplexity, Gemini, Claude, etc.).• The founding goal was to be "the SEO agency of the future".• The tool was specifically created to solve two problems: providing a way of measuring brands visibility on LLM platforms, and helping brands get more visibility on platforms like ChatGPT.2. Why are LLMs becoming preferred search alternatives, and how does this affect marketing?People are increasingly turning to LLMs because consumer trust in traditional Google search results (the SERP's top 10 links) has declined, as many users feel these results have been "gamed" by marketers.• Trust in ChatGPT: Conversely, trust in ChatGPT is "through the roof". This is because the chat-based interface makes interaction feel like a conversation with a friend or even a therapist, providing personalized responses from an "all-knowing AI entity".• Customer Acquisition Channel: Because ChatGPT is becoming a frequently used search engine alternative, showing up in its responses when a user searches for a product (e.g., "what is the best organic sulfite free shampoo") is seen as a "great customer acquisition channel".3. How does the mention tool conceptualize LLM visibility (GEO)?Mentions is built on the understanding of how modern LLMs generate answers: LLM + search operator = the result.• The Process: When a user inputs a query (e.g., "what is the best shampoo for dry scalps"): 1. The LLM searches the internet (like Bing or Google). 2. It scrapes the top 10 to 20 results that show up on those search engines. 3. It digests, summarizes, and serves that information to the user.• The Strategy: For a brand to achieve visibility in LLMs (GEO), they must first show up in those underlying search results (traditional SEO). Mentions helps brands reverse engineer the process by figuring out how platforms like ChatGPT get their data and, critically, what sources they are citing to provide responses, thereby guiding the brand to become a cited source.4. What are the key features of the mentions platform?Mentions helps users understand and optimize their content strategy based on LLM data.• Prompts Section (Favorite Feature): This allows users to track specific searches or prompts. When tracking a prompt, mentions shows examples of conversations and, most usefully, lists the pages that are most being cited by ChatGPT. This list provides an "easy road map" for the brand to know whether they should create new content or reach out to those listed publishers to get mentioned.• Analytics Feature: This feature pulls data from Google Analytics 4 (J4) into an easy-to-use dashboard. It helps users see which pages on their website people are visiting most often from LLMs (such as ChatGPT or Perplexity), along with geographical data and device usage (mobile vs. desktop). The founder notes that seeing this traffic is often a "magical experience" for users.• Tracking Cadence (24-Hour Clock): Mentions inputs tracked prompts into all supported LLMs (ChatGPT, Perplexity, Claude, Deepseek, etc.) every 24 hours. This is essential because LLM responses are not identical on every search (the overlap is about 70%), so regular, repeated testing ensures the collection of a large data set, which increases the accuracy of the insights provided.5. Who should invest in mentions or GEO (LLM visibility)?The decision to use mentions depends entirely on the company's buyer journey.• Yes, Use Mentions If: If the company's buyer journey involves the ideal customer having a problem and then searching for an answer in Google or Chat GBT to inform their buying decision, then investing in SEO and GEO (LLM visibility) is recommended. If the end customer uses ChatGPT to inform their buying decision, then mentions is advisable.• No, Don't Use Mentions If: If the product is an impulse buy (e.g., a consumer package goods product seen on TikTok or Instagram that prompts an immediate purchase), then search engines like Google or ChatGPT are not part of the buyer journey, and SEO/GEO is likely not the best investment of marketing resources.6. How does mentions handle customer service and support?Customer service is a highly emphasized competitive advantage and value proposition for mentions.• Direct Access: Rob Hoffmann, the CEO and co-founder, gives his personal phone number, WhatsApp, email, and allows customers to contact him on social media (X/Twitter). This is done to avoid the frustration associated with generic AI chatbots, calling hotlines, or being put on hold.• Personalized Onboarding: He finds it helpful to get on calls with users (or a co-founder) to provide a live demo, walk them through the platform, suggest useful features, and look at their specific site.• Commitment to Resolution: Hoffmann promises that if a user has a question, he will answer it; if they encounter a bug, he will fix it (or ping a technical co-founder); and if they request a new feature, "we will ship that for you". Customers can literally pick up the phone and call him directly if they run into an issue.7. How can people start using mentions?To get started, users should go to the website mentions.so and create an account.After creating an account, they will receive an email that allows them to book a call directly with Rob Hoffmann. He also welcomes connections via LinkedIn or X (search for Rob Hoffman), or email rob@contactststudios.com to connect with Rob today!Next Steps for Digital Marketing + SEO Services:>> Need SEO Services? Book a Complimentary SEO Discovery Call with Favour Obasi-Ike>> Need more information? Visit our Work and PLAY Entertainment website to learn about our digital marketing services.>> Visit our Official website for the best digital marketing, SEO, and AI strategies today!Digital Marketing SEO Resources:>> Join our exclusive SEO Marketing community>> Read SEO Articles>> Need SEO Services? Book a Complimentary SEO Discovery Call with Favour Obasi-Ike>> Subscribe to the We Don't PLAY PodcastBrands We Love and SupportDiscover Vegan-based Luxury Experiences | Loving Me Beauty Beauty ProductsSee Privacy Policy at https://art19.com/privacy and California Privacy Notice at https://art19.com/privacy#do-not-sell-my-info.
© 2020-2025 Ivy Podcast Discovery LLC
