Podcasts about alphago zero

The machines are coming. Scratch that—they're already here: AIs that propose new combinations of ideas; chatbots that help us summarize texts or write code; algorithms that tell us who to friend or follow, what to watch or read. For a while the reach of intelligent machines may have seemed somewhat limited. But not anymore—or, at least, not for much longer. The presence of AI is growing, accelerating, and, for better or worse, human culture may never be the same.    My guest today is Dr. Iyad Rahwan. Iyad directs the Center for Humans and Machines at the Max Planck Institute for Human Development in Berlin. Iyad is a bit hard to categorize. He's equal parts computer scientist and artist; one magazine profile described him as "the Anthropologist of AI." Labels aside, his work explores the emerging relationships between AI, human behavior, and society. In a recent paper, Iyad and colleagues introduced a framework for understanding what they call "machine culture." The framework offers a way of thinking about the different routes through which AI may transform—is transforming—human culture.    Here, Iyad and I talk about his work as a painter and how he brings AI into the artistic process. We discuss whether AIs can make art by themselves and whether they may eventually develop good taste. We talk about how AIphaGoZero upended the world of Go and about how LLMs might be changing how we speak. We consider what AIs might do to cultural diversity. We discuss the field of cultural evolution and how it provides tools for thinking about this brave new age of machine culture. Finally, we discuss whether any spheres of human endeavor will remain untouched by AI influence.    Before we get to it, a humble request: If you're enjoying the show—and it seems that many of you are—we would be ever grateful if you could let the world know. You might do this by leaving a rating or review on Apple Podcasts, or maybe a comment on Spotify. You might do this by giving us a shout out on the social media platform of your choice. Or, if you prefer less algorithmically mediated avenues, you might do this just by telling a friend about us face-to-face. We're hoping to grow the show and best way to do that is through listener endorsements and word of mouth. Thanks in advance, friends.   Alright, on to my conversation with Iyad Rahwan. Enjoy!   A transcript of this episode will be available soon.   Notes and links 3:00 – Images from Dr. Rahwan's ‘Faces of Machine' portrait series. One of the portraits from the series serves as our tile art for this episode. 11:30 – The “stochastic parrots” term comes from an influential paper by Emily Bender and colleagues. 18:30 – A popular article about DALL-E and the “avocado armchair.” 21:30 – Ted Chiang's essay, “Why A.I. isn't going to make art.” 24:00 – An interview with Boris Eldagsen, who won the Sony World Photography Awards in March 2023 with an image that was later revealed to be AI-generated.  28:30 – A description of the concept of “science fiction science.” 29:00 – Though widely attributed to different sources, Isaac Asimov appears to have developed the idea that good science fiction predicts not the automobile, but the traffic jam.  30:00 – The academic paper describing the Moral Machine experiment. You can judge the scenarios for yourself (or design your own scenarios) here. 30:30 – An article about the Nightmare Machine project; an article about the Deep Empathy project. 37:30 – An article by Cesar Hidalgo and colleagues about the relationship between television/radio and global celebrity. 41:30 – An article by Melanie Mitchell (former guest!) on AI and analogy. A popular piece about that work.   42:00 – A popular article describing the study of whether AIs can generate original research ideas. The preprint is here. 46:30 – For more on AlphaGo (and its successors, AlphaGo Zero and AlphaZero), see here. 48:30 – The study finding that the novel of human Go playing increased due to the influence of AlphaGo. 51:00 – A blogpost delving into the idea that ChatGPT overuses certain words, including “delve.” A recent preprint by Dr. Rahwan and colleagues, presenting evidence that “delve” (and other words overused by ChatGPT) are now being used more in human spoken communication.  55:00 – A paper using simulations to show how LLMs can “collapse” when trained on data that they themselves generated.  1:01:30 – A review of the literature on filter bubbles, echo chambers, and polarization. 1:02:00 – An influential study by Dr. Chris Bail and colleagues suggesting that exposure to opposing views might actually increase polarization.  1:04:30 – A book by Geoffrey Hodgson and Thorbjørn Knudsen, who are often credited with developing the idea of “generalized Darwinism” in the social sciences.  1:12:00 – An article about Google's NotebookLM podcast-like audio summaries. 1:17:3 0 – An essay by Ursula LeGuin on children's literature and the Jungian “shadow.”    Recommendations The Secret of Our Success, Joseph Henrich “Machine Behaviour,” Iyad Rahwan et al.   Many Minds is a project of the Diverse Intelligences Summer Institute, which is made possible by a generous grant from the John Templeton Foundation to Indiana University. The show is hosted and produced by Kensy Cooperrider, with help from Assistant Producer Urte Laukaityte and with creative support from DISI Directors Erica Cartmill and Jacob Foster. Our artwork is by Ben Oldroyd. Our transcripts are created by Sarah Dopierala. Subscribe to Many Minds on Apple, Stitcher, Spotify, Pocket Casts, Google Play, or wherever you listen to podcasts. You can also now subscribe to the Many Minds newsletter here! We welcome your comments, questions, and suggestions. Feel free to email us at: manymindspodcast@gmail.com.  For updates about the show, visit our website or follow us on Twitter (@ManyMindsPod) or Bluesky (@manymindspod.bsky.social).

Welcome to The Nonlinear Library, where we use Text-to-Speech software to convert the best writing from the Rationalist and EA communities into audio. This is: Parameter Scaling Comes for RL, Maybe, published by 1a3orn on January 24, 2023 on LessWrong. TLDR Unlike language models or image classifiers, past reinforcement learning models did not reliably get better as they got bigger. Two DeepMind RL papers published in January 2023 nevertheless show that with the right techniques, scaling up RL model parameters can increase both total reward and sample-efficiency of RL agents -- and by a lot. Return-to-scale has been key for rendering language models powerful and economically valuable; it might also be key for RL, although many important questions remain unanswered. Intro Reinforcement learning models often have very few parameters compared to language and image models. The Vision Transformer has 2 billion parameters. GPT-3 has 175 billion. The slimmer Chinchilla, trained in accord with scaling laws emphasizing bigger datasets, has 70 billion. By contrast, until a month ago, the largest mostly-RL models I knew of were the agents for Starcraft and Dota2, AlphaStar and OpenAI5, which had 139 million and 158 million parameters. And most RL models are far smaller, coming in well under 50 million parameters. The reason RL hasn't scaled up the size of its models is simple -- doing so generally hasn't made them better. Increasing model size in RL can even hurt performance. MuZero Reanalyze gets worse on some tasks as you scale network size. So does a vanilla SAC agent. There has been good evidence for scaling model size in somewhat... non-central examples of RL. For instance, offline RL agents trained from expert examples, such as DeepMind's 1.2-billion parameter Gato or Multi-Game Decision Transformers, clearly get better with scale. Similarly, RL from human feedback on language models generally shows that larger LM's are better. Hybrid systems such as PaLM SayCan benefit from larger language models. But all these cases sidestep problems central to RL -- they have no need to balance exploration and exploitation in seeking reward. In the typical RL setting -- there has generally been little scaling and little evidence for the efficacy of scaling. (Although there has not been no evidence.) None of the above means that the compute spent on RL models is small or that compute scaling does nothing for them. AlphaStar used only a little less compute than GPT-3, and AlphaGo Zero used more, because both of them trained on an enormous number of games. Additional compute predictably improves performance of RL agents. But, rather than getting a bigger brain, almost all RL algorithms spend this compute by (1) training on an enormous number of games (2) or (if concerned with sample-efficiency) by revisiting the games that they've played an enormous number of times. So for a while RL has lacked: (1) The ability to scale up model size to reliably improve performance. (2) (Even supposing the above were around) Any theory like the language-model scaling laws which would let you figure out how to allocate compute between model size / longer training. My intuition is that the lack of (1), and to a lesser degree the lack of (2), is evidence that no one has stumbled on the "right way" to do RL or RL-like problems. It's like language modeling when it only had LSTMS and no Transformers, before the frighteningly straight lines in log-log charts appeared. In the last month, though, two RL papers came out with interesting scaling charts, each showing strong gains to parameter scaling. Both were (somewhat unsurprisingly) from DeepMind. This is the kind of thing that leads me to think "Huh, this might be an important link in the chain that brings about AGI." The first paper is "Mastering Diverse Domains Through World Models", which names its agent DreamerV3. The second is "Human-Timescale Adaptation in an Open-Ended Task Space", which names its agent Adaptive...

Link to original articleWelcome to The Nonlinear Library, where we use Text-to-Speech software to convert the best writing from the Rationalist and EA communities into audio. This is: Parameter Scaling Comes for RL, Maybe, published by 1a3orn on January 24, 2023 on LessWrong. TLDR Unlike language models or image classifiers, past reinforcement learning models did not reliably get better as they got bigger. Two DeepMind RL papers published in January 2023 nevertheless show that with the right techniques, scaling up RL model parameters can increase both total reward and sample-efficiency of RL agents -- and by a lot. Return-to-scale has been key for rendering language models powerful and economically valuable; it might also be key for RL, although many important questions remain unanswered. Intro Reinforcement learning models often have very few parameters compared to language and image models. The Vision Transformer has 2 billion parameters. GPT-3 has 175 billion. The slimmer Chinchilla, trained in accord with scaling laws emphasizing bigger datasets, has 70 billion. By contrast, until a month ago, the largest mostly-RL models I knew of were the agents for Starcraft and Dota2, AlphaStar and OpenAI5, which had 139 million and 158 million parameters. And most RL models are far smaller, coming in well under 50 million parameters. The reason RL hasn't scaled up the size of its models is simple -- doing so generally hasn't made them better. Increasing model size in RL can even hurt performance. MuZero Reanalyze gets worse on some tasks as you scale network size. So does a vanilla SAC agent. There has been good evidence for scaling model size in somewhat... non-central examples of RL. For instance, offline RL agents trained from expert examples, such as DeepMind's 1.2-billion parameter Gato or Multi-Game Decision Transformers, clearly get better with scale. Similarly, RL from human feedback on language models generally shows that larger LM's are better. Hybrid systems such as PaLM SayCan benefit from larger language models. But all these cases sidestep problems central to RL -- they have no need to balance exploration and exploitation in seeking reward. In the typical RL setting -- there has generally been little scaling and little evidence for the efficacy of scaling. (Although there has not been no evidence.) None of the above means that the compute spent on RL models is small or that compute scaling does nothing for them. AlphaStar used only a little less compute than GPT-3, and AlphaGo Zero used more, because both of them trained on an enormous number of games. Additional compute predictably improves performance of RL agents. But, rather than getting a bigger brain, almost all RL algorithms spend this compute by (1) training on an enormous number of games (2) or (if concerned with sample-efficiency) by revisiting the games that they've played an enormous number of times. So for a while RL has lacked: (1) The ability to scale up model size to reliably improve performance. (2) (Even supposing the above were around) Any theory like the language-model scaling laws which would let you figure out how to allocate compute between model size / longer training. My intuition is that the lack of (1), and to a lesser degree the lack of (2), is evidence that no one has stumbled on the "right way" to do RL or RL-like problems. It's like language modeling when it only had LSTMS and no Transformers, before the frighteningly straight lines in log-log charts appeared. In the last month, though, two RL papers came out with interesting scaling charts, each showing strong gains to parameter scaling. Both were (somewhat unsurprisingly) from DeepMind. This is the kind of thing that leads me to think "Huh, this might be an important link in the chain that brings about AGI." The first paper is "Mastering Diverse Domains Through World Models", which names its agent DreamerV3. The second is "Human-Timescale Adaptation in an Open-Ended Task Space", which names its agent Adaptive...

Welcome to The Nonlinear Library, where we use Text-to-Speech software to convert the best writing from the Rationalist and EA communities into audio. This is: Embedding safety in ML development, published by Chin Ze Shen on October 31, 2022 on The AI Alignment Forum. This post was written as part of Refine. Thanks to Adam Shimi, Alexander Gietelink Oldenziel, and Vanessa Kosoy for helpful discussion and feedback. Summary This post aims to: Advocate for embedding safety into development of machine learning models Propose a framing on how to think about safety, where three factors contribute to an AI being dangerous, namely misalignment, optimization, and influence Discuss the pros and cons of this framing None of the ideas in this post are novel, and some of them may not even be practical, but they may be a useful framing to the problem of AI safety. Introduction Motivations There are many ways to get to AGI. It could be recursively self improving AI, or something more brain-like, or something that comes from improvements in machine learning systems, or something else much more alien to us. I don't know which pathways are the ones our world will eventually end up in, but given recent rapid advances in ML recently, in this post I will address the scenario where the first dangerous AGI is achieved by mostly scaling up and improving upon present-day ML systems. In the absence of a robust solution to the hard problem of alignment, I think we should at least try to make the development of machine learning models go safely for as long as possible. This may be especially valuable if the first AGI we get is not yet an all-powerful superintelligent. It might be a fairly weak AGI capable of inflicting a lot of damage, but the damage could have been prevented. Not having a solution to the hard problem should not mean getting ourselves into preventable disasters. What safety means in this context Most safety fields in the real world don't solve safety with guarantees. Whether it is flying an aircraft, operating a nuclear plant, or constructing a building, safety is all about effective risk management. This framing of safety has not been commonly used in the field of AI safety, for very good reasons, as the problem of a misaligned AGI is an unbounded one, where a ‘safety failure' with a strong form of AGI could mean human extinction even with just a tiny degree of misalignment. For the purpose of this post, I am using the word “safety” in a more narrow sense than most people would. In terms of time-scale, I refer to “we are safe for now” rather than “things go great in the long run”; while in terms of failure mode, I refer to “we are safe from misaligned AIs” rather than “we are safe from AIs being used by bad actors and multipolar failures etc”. Framing the problem The question to be addressed is: what makes an AI harmful? I propose a framing that contains three ingredients: Misalignment at any level Optimization towards some goal Influence over the world The combination of optimization and influence can be loosely thought of as agency. Without any of the above ingredients, there will not be a catastrophe caused by a misaligned ‘agentic' AI, as there would only be the following kinds of AI: Aligned: A fully aligned AI that has strong optimization and high influence over the world will do exactly what we want effectively, possibly leading us to utopia. Dormant: A misaligned AI that has a lot of influence over the world but not trying to do anything will probably not really be doing anything, like a deactivated nuclear bomb. Sponge: A misaligned AI that has a lot of optimization power but no influence over the world will not be doing anything in the real world, effectively being a sponge AI. I consider most of the powerful AIs today, such as AlphaGo Zero and DALL-E 2, as sponges for now, as they neither construct world-models nor have physical control over things in the physical world, but it is debatable if they will remain so in the fu...

Welcome to The Nonlinear Library, where we use Text-to-Speech software to convert the best writing from the Rationalist and EA communities into audio. This is: DeepMind's generalist AI, Gato: A non-technical explainer, published by frances lorenz on May 16, 2022 on LessWrong. Summary DeepMind's recent paper, A Generalist Agent, catalyzed a wave of discourse regarding the speed at which current artificial intelligence systems are improving and the risks posed by these increasingly advanced systems. We aim to make Gato accessible to non-technical folks by: (i) providing a non-technical summary, and (ii) discussing the relevant implications related to existential risk and AI policy. Introduction DeepMind has just introduced its new agent, Gato: the most general machine learning (ML) model to date. If you're familiar with arguments for the potential risks posed by advanced AI systems, you'll know the term general carries strong implications. Today's ML systems are advancing quickly; however, even the best systems we see are narrow in the tasks they can accomplish. For example, DALL-E impressively generates images that rival human creativity; however, it doesn't do anything else. Similarly, large language models like GPT-3 perform well on certain text-based tasks, like sentence completion, but poorly on others, such as arithmetic (Figure 1). If future AI systems are to exhibit human-like intelligence, they'll need to use various skills and information to complete diverse tasks across different contexts. In other words, they'll need to exhibit general intelligence in the same way humans do—a type of system broadly referred to as artificial general intelligence (AGI). While AGI systems could lead to hugely positive innovations, they also have the potential to surpass human intelligence and become “superintelligent”. If a superintelligent system were unaligned, it could be difficult or even impossible to control for and predict its behavior, leaving humans vulnerable. Figure 1: An attempt to teach GPT-3 addition. The letter ‘Q' denotes human input while ‘A' denotes GPT-3's response (from Peter Wildeford's tweet) So what exactly has DeepMind created? Gato is a single neural network capable of performing hundreds of distinct tasks. According to DeepMind, it can, “play Atari, caption images, chat, stack blocks with a real robot arm and much more, deciding based on its context whether to output text, joint torques, button presses, or other tokens.” It's not currently analogous to human-like intelligence; however, it does exhibit general capabilities. In the rest of this post, we'll provide a non-technical summary of DeepMind's paper and explore: (i) what this means for potential future existential risks posed by advanced AI and (ii) some relevant AI policy considerations. A Summary of Gato How was Gato built? The technique used to train Gato is slightly different from other famous AI agents. For example, AlphaGo, the AI system that defeated world champion Go player Lee Sedol in 2016, was trained largely using a sophisticated form of trial and error called reinforcement learning (RL). While the initial training process involved some demonstrations from expert Go players, the next iteration named AlphaGo Zero removed these entirely, mastering games solely by playing itself. By contrast, Gato was trained to imitate examples of “good” behavior in 604 distinct tasks. These tasks include: Simulated control tasks, where Gato has to control a virtual body in a simulated environment. Vision and language tasks, like labeling images with corresponding text captions. Robotics, specifically the common RL task of stacking blocks. Examples of good behavior were collected in a few different ways. For simulated control and robotics, examples were collected from other, more specialized AI agents trained using RL. For vision and language tasks, “behavior” took the form of text and images generated by humans, largely scraped from the web. Results Control ...

Link to original articleWelcome to The Nonlinear Library, where we use Text-to-Speech software to convert the best writing from the Rationalist and EA communities into audio. This is: DeepMind's generalist AI, Gato: A non-technical explainer, published by frances lorenz on May 16, 2022 on LessWrong. Summary DeepMind's recent paper, A Generalist Agent, catalyzed a wave of discourse regarding the speed at which current artificial intelligence systems are improving and the risks posed by these increasingly advanced systems. We aim to make Gato accessible to non-technical folks by: (i) providing a non-technical summary, and (ii) discussing the relevant implications related to existential risk and AI policy. Introduction DeepMind has just introduced its new agent, Gato: the most general machine learning (ML) model to date. If you're familiar with arguments for the potential risks posed by advanced AI systems, you'll know the term general carries strong implications. Today's ML systems are advancing quickly; however, even the best systems we see are narrow in the tasks they can accomplish. For example, DALL-E impressively generates images that rival human creativity; however, it doesn't do anything else. Similarly, large language models like GPT-3 perform well on certain text-based tasks, like sentence completion, but poorly on others, such as arithmetic (Figure 1). If future AI systems are to exhibit human-like intelligence, they'll need to use various skills and information to complete diverse tasks across different contexts. In other words, they'll need to exhibit general intelligence in the same way humans do—a type of system broadly referred to as artificial general intelligence (AGI). While AGI systems could lead to hugely positive innovations, they also have the potential to surpass human intelligence and become “superintelligent”. If a superintelligent system were unaligned, it could be difficult or even impossible to control for and predict its behavior, leaving humans vulnerable. Figure 1: An attempt to teach GPT-3 addition. The letter ‘Q' denotes human input while ‘A' denotes GPT-3's response (from Peter Wildeford's tweet) So what exactly has DeepMind created? Gato is a single neural network capable of performing hundreds of distinct tasks. According to DeepMind, it can, “play Atari, caption images, chat, stack blocks with a real robot arm and much more, deciding based on its context whether to output text, joint torques, button presses, or other tokens.” It's not currently analogous to human-like intelligence; however, it does exhibit general capabilities. In the rest of this post, we'll provide a non-technical summary of DeepMind's paper and explore: (i) what this means for potential future existential risks posed by advanced AI and (ii) some relevant AI policy considerations. A Summary of Gato How was Gato built? The technique used to train Gato is slightly different from other famous AI agents. For example, AlphaGo, the AI system that defeated world champion Go player Lee Sedol in 2016, was trained largely using a sophisticated form of trial and error called reinforcement learning (RL). While the initial training process involved some demonstrations from expert Go players, the next iteration named AlphaGo Zero removed these entirely, mastering games solely by playing itself. By contrast, Gato was trained to imitate examples of “good” behavior in 604 distinct tasks. These tasks include: Simulated control tasks, where Gato has to control a virtual body in a simulated environment. Vision and language tasks, like labeling images with corresponding text captions. Robotics, specifically the common RL task of stacking blocks. Examples of good behavior were collected in a few different ways. For simulated control and robotics, examples were collected from other, more specialized AI agents trained using RL. For vision and language tasks, “behavior” took the form of text and images generated by humans, largely scraped from the web. Results Control ...

Welcome to The Nonlinear Library, where we use Text-to-Speech software to convert the best writing from the Rationalist and EA communities into audio. This is: DeepMind's generalist AI, Gato: A non-technical explainer, published by frances lorenz on May 16, 2022 on The Effective Altruism Forum. Summary DeepMind's recent paper, A Generalist Agent, catalyzed a wave of discourse regarding the speed at which current artificial intelligence systems are improving and the risks posed by these increasingly advanced systems. We aim to make this paper accessible to non-technical folks by: (i) providing a non-technical summary, and (ii) discussing the relevant implications related to existential risk and AI policy. Introduction DeepMind has just introduced its new agent, Gato: the most general machine learning (ML) model to date. If you're familiar with arguments for the potential risks posed by advanced AI systems, you'll know the term general carries strong implications. Today's ML systems are advancing quickly; however, even the best systems we see are narrow in the tasks they can accomplish. For example, DALL-E impressively generates images that rival human creativity; however, it doesn't do anything else. Similarly, large language models like GPT-3 perform well on certain text-based tasks, like sentence completion, but poorly on others, such as arithmetic (Figure 1). If future AI systems are to exhibit human-like intelligence, they'll need to use various skills and information to complete diverse tasks across different contexts. In other words, they'll need to exhibit general intelligence in the same way humans do—a type of system broadly referred to as artificial general intelligence (AGI). While AGI systems could lead to hugely positive innovations, they also have the potential to surpass human intelligence and become “superintelligent”. If a superintelligent system were unaligned, it could be difficult or even impossible to control for and predict its behavior, leaving humans vulnerable. Figure 1: An attempt to teach GPT-3 addition. The letter ‘Q' denotes human input while ‘A' denotes GPT-3's response (from Peter Wildeford's tweet) So what exactly has DeepMind created? Gato is a single neural network capable of performing hundreds of distinct tasks. According to DeepMind, it can, “play Atari, caption images, chat, stack blocks with a real robot arm and much more, deciding based on its context whether to output text, joint torques, button presses, or other tokens.” It's not currently analogous to human-like intelligence; however, it does exhibit general capabilities. In the rest of this post, we'll provide a non-technical summary of DeepMind's paper and explore: (i) what this means for potential future existential risks posed by advanced AI and (ii) some relevant AI policy considerations. A Summary of Gato How was Gato built? The technique used to train Gato is slightly different from other famous AI agents. For example, AlphaGo, the AI system that defeated world champion Go player Lee Sedol in 2016, was trained largely using a sophisticated form of trial and error called reinforcement learning (RL). While the initial training process involved some demonstrations from expert Go players, the next iteration named AlphaGo Zero removed these entirely, mastering games solely by playing itself. By contrast, Gato was trained to imitate examples of “good” behavior in 604 distinct tasks. These tasks include: Simulated control tasks, where Gato has to control a virtual body in a simulated environment. Vision and language tasks, like labeling images with corresponding text captions. Robotics, specifically the common RL task of stacking blocks. Examples of good behavior were collected in a few different ways. For simulated control and robotics, examples were collected from other, more specialized AI agents trained using RL. For vision and language tasks, “behavior” took the form of text and images generated by humans, largely scraped from ...

Welcome to The Nonlinear Library, where we use Text-to-Speech software to convert the best writing from the Rationalist and EA communities into audio. This is Iterated Amplification, Part 14: AlphaGo Zero and capability amplification, published by paulfchristiano. Crossposted from the AI Alignment Forum. May contain more technical jargon than usual. AlphaGo Zero is an impressive demonstration of AI capabilities. It also happens to be a nice proof-of-concept of a promising alignment strategy. How AlphaGo Zero works AlphaGo Zero learns two functions (which take as input the current board): A prior over moves p is trained to predict what AlphaGo will eventually decide to do. A value function v is trained to predict which player will win (if AlphaGo plays both sides) Both are trained with supervised learning. Once we have these two functions, AlphaGo actually picks it moves by using 1600 steps of Monte Carlo tree search (MCTS), using p and v to guide the search. It trains p to bypass this expensive search process and directly pick good moves. As p improves, the expensive search becomes more powerful, and p chases this moving target. Iterated capability amplification In the simplest form of iterated capability amplification, we train one function: A “weak” policy A, which is trained to predict what the agent will eventually decide to do in a given situation. Just like AlphaGo doesn't use the prior p directly to pick moves, we don't use the weak policy A directly to pick actions. Instead, we use a capability amplification scheme: we call A many times in order to produce more intelligent judgments. We train A to bypass this expensive amplification process and directly make intelligent decisions. As A improves, the amplified policy becomes more powerful, and A chases this moving target. In the case of AlphaGo Zero, A is the prior over moves, and the amplification scheme is MCTS. (More precisely: A is the pair (p, v), and the amplification scheme is MCTS + using a rollout to see who wins.) Outside of Go, A might be a question-answering system, which can be applied several times in order to first break a question down into pieces and then separately answer each component. Or it might be a policy that updates a cognitive workspace, which can be applied many times in order to “think longer” about an issue. The significance Reinforcement learners take a reward function and optimize it; unfortunately, it's not clear where to get a reward function that faithfully tracks what we care about. That's a key source of safety concerns. By contrast, AlphaGo Zero takes a policy-improvement-operator (like MCTS) and converges towards a fixed point of that operator. If we can find a way to improve a policy while preserving its alignment, then we can apply the same algorithm in order to get very powerful but aligned strategies. Using MCTS to achieve a simple goal in the real world wouldn't preserve alignment, so it doesn't fit the bill. But “think longer” might. As long as we start with a policy that is close enough to being aligned — a policy that “wants” to be aligned, in some sense — allowing it to think longer may make it both smarter and more aligned. I think designing alignment-preserving policy amplification is a tractable problem today, which can be studied either in the context of existing ML or human coordination. So I think it's an exciting direction in AI alignment. A candidate solution could be incorporated directly into the AlphaGo Zero architecture, so we can already get empirical feedback on what works. If by good fortune powerful AI systems look like AlphaGo Zero, then that might get us much of the way to an aligned AI. Thanks for listening. To help us out with The Nonlinear Library or to learn more, please visit nonlinear.org.

Link to original articleWelcome to The Nonlinear Library, where we use Text-to-Speech software to convert the best writing from the Rationalist and EA communities into audio. This is Iterated Amplification, Part 14: AlphaGo Zero and capability amplification, published by paulfchristiano. Crossposted from the AI Alignment Forum. May contain more technical jargon than usual. AlphaGo Zero is an impressive demonstration of AI capabilities. It also happens to be a nice proof-of-concept of a promising alignment strategy. How AlphaGo Zero works AlphaGo Zero learns two functions (which take as input the current board): A prior over moves p is trained to predict what AlphaGo will eventually decide to do. A value function v is trained to predict which player will win (if AlphaGo plays both sides) Both are trained with supervised learning. Once we have these two functions, AlphaGo actually picks it moves by using 1600 steps of Monte Carlo tree search (MCTS), using p and v to guide the search. It trains p to bypass this expensive search process and directly pick good moves. As p improves, the expensive search becomes more powerful, and p chases this moving target. Iterated capability amplification In the simplest form of iterated capability amplification, we train one function: A “weak” policy A, which is trained to predict what the agent will eventually decide to do in a given situation. Just like AlphaGo doesn't use the prior p directly to pick moves, we don't use the weak policy A directly to pick actions. Instead, we use a capability amplification scheme: we call A many times in order to produce more intelligent judgments. We train A to bypass this expensive amplification process and directly make intelligent decisions. As A improves, the amplified policy becomes more powerful, and A chases this moving target. In the case of AlphaGo Zero, A is the prior over moves, and the amplification scheme is MCTS. (More precisely: A is the pair (p, v), and the amplification scheme is MCTS + using a rollout to see who wins.) Outside of Go, A might be a question-answering system, which can be applied several times in order to first break a question down into pieces and then separately answer each component. Or it might be a policy that updates a cognitive workspace, which can be applied many times in order to “think longer” about an issue. The significance Reinforcement learners take a reward function and optimize it; unfortunately, it's not clear where to get a reward function that faithfully tracks what we care about. That's a key source of safety concerns. By contrast, AlphaGo Zero takes a policy-improvement-operator (like MCTS) and converges towards a fixed point of that operator. If we can find a way to improve a policy while preserving its alignment, then we can apply the same algorithm in order to get very powerful but aligned strategies. Using MCTS to achieve a simple goal in the real world wouldn't preserve alignment, so it doesn't fit the bill. But “think longer” might. As long as we start with a policy that is close enough to being aligned — a policy that “wants” to be aligned, in some sense — allowing it to think longer may make it both smarter and more aligned. I think designing alignment-preserving policy amplification is a tractable problem today, which can be studied either in the context of existing ML or human coordination. So I think it's an exciting direction in AI alignment. A candidate solution could be incorporated directly into the AlphaGo Zero architecture, so we can already get empirical feedback on what works. If by good fortune powerful AI systems look like AlphaGo Zero, then that might get us much of the way to an aligned AI. Thanks for listening. To help us out with The Nonlinear Library or to learn more, please visit nonlinear.org.

Sangat jarang saya bisa bertemu dan diskusi dengan orang yang mendalami bidang Reinforcement Learning. Muhammad Arrasy Rahman (@raharrasy93) mungkin adalah salah satu yang jadi referensi saya kalau ada pertanyaan seputar bidang yang jadi sorotan orang-orang berkat AlphaGo, AlphaGo Zero, dan turunannya dalam beberapa tahun terakhir. Episode kali ini membahas perkembangan di bidang RL dan kemungkinan implementasinya di dunia nyata. Diskusi ini juga menyinggung hubungan RL dengan game theory, supply chain, dan bidang-bidang lain dari segi teori dan implementasi.

DeepMind’ın AlphaGo programı, antik bir masa oyunu olan Go'da insan rakibini yendiğinde bu program, yapay zeka alanında büyük bir sıçrama yaratmıştı. AlphaGo önceden yazılmış bir dizi talimat vasıtasıyla değil de pratik ve geri bildirimler yoluyla eğitilmişti. Bu durum, yeni nesil makine öğrenme teknolojileri… Seslendiren: Altay Kenger

Welcome to another special edition of „Mediocrity and Madness“! Usually this Podcast is dedicated to the ever-widening gap between talk and reality in our big organizations, most notably in our global corporates. Well, I might have to admit that in some cases the undertone is a tiny bit angry and another bit tongue-in-cheek. The title might indicate that. Today’s episode is not like this. Well, it is but in a different way. Upon reflection, it still addresses a mighty chasm between talk and reality but the reason for this chasm appears more forgivable to me than those many dysfunctions we appear to have accepted against better judgement. Today’s podcast is about artificial intelligence and our struggles to put it to use in businesses. This podcast is to some measure inspired by what I learned in and around two programs of Allianz, “IT Literacy for top executives” and “AI for the business”, which I had the privilege and the pleasure to help developing and facilitating. I am tempted to begin this episode with the same claim I used in the last (German) one: With artificial intelligence it is like with teenage sex. Everybody talks about it, but nobody really knows how it works. Everybody thinks that everyone else does it. Thus, everybody claims he does it. And again, Dan Ariely gets all the credits for coining that phrase with “Big Data” instead of “artificial intelligence” which is actually a bit related anyway. Or not. As we will see later. To begin with, the big question is: What is “artificial intelligence” after all? The straightforward way to answering that question is to first define what intelligence is in general and then apply the notion that “artificial” is just when the same is done by machines. Yet here begins the problem. There simply is no proper definition of intelligence. Some might say, intelligence is what discerns man from animal but that’s not very helpful, too. Where’s the boarder. When I was a boy, I read that a commonplace definition was that humans use tools while animals don’t. Besides the question whether that little detail would be one that made us truly proud of our human intelligence, multiple examples of animals using tools have been found since. To make a long story short, there is no proper and general definition of intelligence. Thus, we end up with some self-referentiality: “It’s intelligent if it behaves like a human”. In a way, that’s quite a dissatisfying definition, most of all because it leaves no room for types of intelligences that behave – or “are” – significantly non-human. “Black swan” is greeting. But we’re detouring into philosophy. Back to our problem at hand: What is artificial intelligence after all? Well, if it’s intelligent, if it behaves like a human, then the logical answer to this question is: “artificial intelligence is when a computer/machine behaves like a human”. For practical purposes this is something we can work with. Yet even then another question looms: How do we evaluate whether it behaves like a human? Being used to some self-referentiality already, the answer is quite straight forward: “It behaves like a human if other humans can’t tell the difference from human behavior.” This is actually the essence of what is called the “Turing test”, devised by the famous British mathematician Alan Turing who next to basically inventing what we today call computer sciences helped solving the Enigma encryption during World War II. Turing’s biography is as inspiring as it is tragic and I wouldn’t mind if you stopped listening to this humble podcast and explored Turing in a bit more depth, for example by watching “The imitation game” starring Benedict Cumberbatch. If you decide to stay with me instead of Cumberbatch, that’s where we finally are: “Artificial intelligence is when a machine/robot behaves in a way that humans can’t discern that behavior from human behavior.” As you might imagine, the respective tests have to be designed properly so that biases are avoided. And, of course, also the questions or problems designed to ascertain human or less human behavior have to be designed carefully. These are subjects of more advanced versions of the Turing test but in the end, the ultimate condition remains the same: A machine is regarded intelligent if it behaves like a human. (Deliberately) stupid? It has taken us some time to establish this somewhat flawed, extremely human-centric but workable definition of machine intelligence. It poses some questions and it helps answering some others. One question that is discussed around the Turing test is indeed whether would-be artificial intelligences should deliberately put a few mistakes into their behavior even despite better knowledge, just in order to appear more human. I think that question comes more from would-be philosophers than it is a serious one to consider. Yet, you could argue that if taking the Turing test seriously, in order to convince a human of being a fellow human the occasional mistake is appropriate. After all, “to err is human”. Again, the question appears a bit stupid to me. Would you really argue that it is intelligent only if it occasionally errs? The other side of that coin though is quite relevant. In many discussions about machine intelligence, the implicit or explicit requirement appears to be: If it’s done by a machine, it needs to be 100%. I reason that’s because when dealing with computer algorithms, like calculating for example the trajectory of a moon rocket, we’re used to zero errors; given that the programming is right, that there are no strange glitches in the hardware and that the input data isn’t faulty as such. Writing that, a puzzling thought enters my mind: We trustin machine perfection and expect human imperfection. Not a good outlook in regard to human supremacy. Sorry, I’m on another detour. Time to get back to the question of intelligence. If we define intelligence as behavior being indiscernible from human one, why then do we wonder if machine intelligence doesn’t yield 100% perfect results. Well, for the really complex problems it would actually be impossible to define what “100% perfect” even is, neither ex ante nor ex post but let’s stick to the simpler problems for now: pattern recognition, predictive analysis, autonomous driving … . Intelligent beings make mistakes. Even those whose intelligence is focused onto a specific task. Human radiologists identify some spots on their pictures falsely as positive signs of cancer whilst they overlook others that actually would be malicious. So do machines trained to the same purpose. Competition I am rather sure that the kind listener’s intuitive reaction at this point is: “Who cares? – If the machine makes less errors than her human counterpart, let her take the lead!” And of course, this is the only logical conclusion. Yet quite often, here’s one major barrier to embracing artificial intelligence. Our reaction to machines threatening to become better than us but not totally perfect is poking for the outliers and inflating them until the use of machine intelligence feels somewhat disconcerting. Well, they are competitors after all, aren’t they? The radiologist case is especially illuminating. In fact, the problem is that amongst human radiologists there is a huge, huge spread in competency. Whilst a few radiologists are just brilliant in analyzing their pictures, others are comparatively poor. The gap not only results from experience or attitude, there are also significant differences from county to country for example. Thus, even if the machine would not beat the very best of radiologists, it would be a huge step ahead and saving many, many lives if one could just provide a better average across the board;  – which is what commonly available machines geared to the task do. Guess what your average radiologist thinks about that. – Ah, and don’t mind, if the machine would not yet be better than her best human colleagues, it is but a matter of weeks or months or maybe a year or two until she is as we will see in a minute. You still don’t believe that this impedes the adaption of artificial intelligence? – Look this example that made it into the feuilletons not long ago. Autonomous driving. Suppose you’re sitting in a car that is driven autonomously by some kind of artificial intelligence. All of a sudden, another car – probably driven by a human intelligence – comes towards you on the rather narrow street you’re driven through. Within microseconds, your car recognizes its choices: divert to the right and kill a group of kids playing there, divert to the left and kill some adults in their sixties one of which it recognizes as an important advisor to an even more important politician or keep the track and kill both, the occupants of the oncoming car … and unfortunately you yourself. The dilemma has been stylized to a kind of fundamental question by some would-be philosophers with the underlying notion of “if we can’t solve that dilemma rationally, we might better give up the whole idea of autonomous driving for good.” Well, I am exaggerating again but there is some truth in that. Now, as the dilemma is inextricable as such: bye, bye autonomous driving! Of course, the real answer is all but philosophical. Actually, it doesn’t matter what choice the intelligence driving our car makes. It might actually just throw a dice in its random access memory. We have thousands of traffic victims every year anyway. Humankind has decided to live with that sad fact as the advantages of mobility outweigh these bereavements. We have invented motor liability insurance exactly for that reason. Thus, the only and very pragmatic question has to be: Do the advantages of autonomous driving outweigh some sad accidents? – And fortunately, probability is that autonomous driving will massively reduce the number of traffic accidents so the question is actually a very simple one to deal with. Except probably for motor insurance companies … and some would-be philosophers. Irreversible Here’s another intriguing thing with artificial intelligence: irreversibility. As soon as machine intelligence has become better than man in a specific area, the competition is won forever by the machines. Or lost for humankind. Simple: as soon as your artificial radiologist beats her human colleague, the latter one will never catch up again. On the contrary. The machine will improve further, in some cases very fast. Man might improve a little, over time but by far not at the same speed as his silicon colleague … or competitor … or potential replacement. In some cases, the world splits into two parallel ones: the machine world and the human world. This is what happened in 1997 with the game of Chess when Deep Blue beat the then world champion Gary Kasparow. Deep Blue wasn’t even an intelligence. It was just a brute force with input from some chess savvy programmers but then humans have lost the game to the machines, forever. In today’s chess tournaments not the best players on earth compete but the best human players. They might use computers to improve their game but none of them would stand the slightest chance against a halfway decent artificial chess intelligence … or even a brute force algorithm. The loss of chess for humankind is a rather ancient story compared to the game of Go. Go being multitudes more complex than chess resisted the machines about twenty years more. Brute force doesn’t work for Go and thus it took until 2016 until AlphaGo, an artificial intelligence designed to play Go by Google’s DeepMind finally conquered that stronghold of humanity. That year, AlphaGo defeated Lee Sedol, one of the best players in the world. A few months later, the program also defeated Ke Jie, the then top-ranking player in the world. Most impressive though it is that again only a few months later DeepMind published another version of its Go-genius: AlphaGo Zero. Whilst AlphaGo had been trained with huge numbers of Go matches played by human players, AlphaGo Zero had to be taught only the rules of the game and developed its skills purely by playing against versions of itself. After three days, this version beat her predecessor that had won against Lee Sedol 100:0. And again only three months later, another version was deployed. AlphaZero learnt the games of Chess and Go and Shogi, another highly complex strategy game, in only a few hours and defeated all previous versions in a sweep. By then, man was out of the picture for what can be considered an eternity by measures of AI development cycles. AlphaZero not only plays a better Go – or Chess – than any human does, it develops totally new strategies and tactics to play the game, it plays moves never considered reasonable before by its carbon-based predecessors. It has transcended its creators in the game and never again will humanity regain that domain. This, you see, is the nature of artificial intelligence: as soon as it has gained superiority in a certain domain, this domain is forever lost for humankind. If anything, another technology will surpass its predecessor. We and our human brains won’t. We might comfort ourselves that it’s only rather mundane tasks that we cede to machines of specialized intelligence, that it’s a long way still towards a more universal artificial intelligence and that after all, we’re the creators of these intelligences … . But the games of Chess and Go are actually not quite so mundane and the development is somewhat exponential. Finally, a look into ancient mythology is all but comforting. Take Greece as an example: the progenitor of gods, Uranos, was emasculated by his offspring, the Titans and these again were defeated and punished by their offspring, the Olympians, who then ruled the world, most notably Zeus, Uranos’ grandson. Well, Greek mythology is probably not what the kind listener expects from a podcast about artificial intelligence. Hence, back to business. AI is not necessarily BIG Data Here’s a not so uncommon misconception: AI or advanced analytics is always Big Data or – more exactly: Big Data is a necessary prerequisite for advanced analytics. We could make use of the AlphaZero example again. There could hardly be less data necessary. Just a few rules of the game and off we go! “Wait”, some will argue, “our business problems aren’t like this. What we want is predictive analysis and that’s Big Data for sure!”. I personally and vehemently believe this is a misconception. I actually assume, it is a misconception with a purpose but before sinking deeper into speculation, let’s look at an example, a real business problem. I have spent quite some years in the insurance business. Hence please apologize for me using an insurance example. It is very simple. The idea is using artificial intelligence for calculating insurance premiums, specifically motor insurance third party liability (TPL). Usually, this is a mandatory insurance. The risk it covers is that you in your capacity of driving a car – or parking it – damage an object that belongs to someone else or that you injure someone else. Usually, your insurance premium should reflect the risk you want to cover. Thus, in the case of TPL the essential question from an actuary’s point of view is the following one: Is the person under inspection a good driver or a not so good one? “Good” in the insurer’s sense: less prone to cause an accident and if so, one that usually doesn’t come with a big damage. There are zillions of ways to approach that problem. The best would probably be to get an individual psychological profile of the respective person, add a decently detailed analysis of her driving patterns (where, when, …) and calculate the premium based on that analysis, maybe using some sort of artificial intelligence in order to cope with the complex set of data. The traditional way is comparatively simplistic and indirect. We use a mere handful of data, some of them related to the car like type and registration code, some personal data like age or homeownership and some about driving patterns, mostly yearly mileage and calculate a premium out of these few by some rather simple statistical analysis. If we were looking for more Big Data-ish solutions we could consider basing our calculation on social media timelines. Young males posting photos that show them Friday and Saturday nights in distant clubs with fancy drinks in their hands should emerge with way higher premiums than their geeky contemporaries who spend their weekends in front of some computers using their cars only to drive to the next fast food restaurant or once a week to the comic book shop. The shades in between might be subtle and an artificial intelligence might come up with some rather delicate distinctions. And you might not even need a whole timeline. Just one picture might suffice. The forms of our faces, our haircut, the glasses we fancy, the jewelry we wear, the way we twinkle our noses … might well be very good indicators of our driving behavior. Definitely a job for an artificial intelligence. I’m sure, you can imagine other avenues. Some are truly Big Data, others are rather small in terms of data … and fancy learning machines. The point is, these very different approaches may well yield very similar results ie, a few data related to your car might reveal quite as much about the question at hand as an analysis of your Instagram story. The fundamental reason is that data as such are worthless. Valuable is only what we extract from that data. This is the so-called DIKW hierarchy. Data, Information, Knowledge, Wisdom. The true challenge is extracting wisdom from data. And the rule is not: more data – more wisdom. On the contrary. Too much data might in fact clutter the way to wisdom. And in any case, very different data might represent the same information, knowledge or wisdom. As what concerns our example, I have first of all to admit that I have nor analytical proof – or wisdom – about specifics I am going to discuss but I feel confident that the examples illustrate the point. Here we go. The type of car – put into in the right correlation with a few other data -- might already contain most of the knowledge you could gain from a full-blown psychological analysis or a comprehensive inspection of a person’s social media profile. Data representing a 19 year old male, living in a certain area of town, owning a used but rather high powered car, driving a certain mileage per year might very well contain the same information with respect to our question about “good” driving as all the pictures we find in his Facebook timeline. And the other way around. The same holds true for the information we might get out of a single static photo. Yet the Facebook timeline or the photo are welling over with information that is irrelevant for our specific problem. Or irrelevant at all. And it is utterly difficult to get a) the necessary data in a proper breadth and quality at all and b) to distill relevant information, knowledge and wisdom from this cornucopia of data.  Again: more data does not necessarily mean more wisdom! It might. But one kind of data might – no: will – contain the same information as other kinds. Even the absence of data might contain information or knowledge. Assume for instance, you have someone explicitly denying her consent to using her data for marketing purposes. That might mean she is anxious about her data privacy which in turn might indicate that she is also concerned about other burning social and environmental issues which then might indicate she doesn’t use her car a lot and if so in a rather responsible way … . You get the point. Most probably that whole chain of reasoning won’t work having that single piece of data in isolation but put into the context of other data there might actually be wisdom. Actually, looking at the whole picture, this might not even be a chain of reasoning but more a description of the certain state of things that denies decomposition into human logic. Which leads us to another issue with artificial intelligence. The unboxing problem Artificial intelligences, very much like their human contemporaries, can’t always be understood easily. That is, the logic, the chain of reasoning, the parameters that causally determine certain outcomes, decisions or predictions are in many cases less than transparent. At the same time, we humans demand from artificial intelligence what we can’t deliver for our own reasoning: this very transparency. Quite like us demanding 100% machine perfection, some control-instinct of ours claims: If it’s not transparent to us (humans), it isn’t worth much. Hence, a line of research in the field of artificial intelligence has developed: “Unboxing the AI”. Except for some specific cases yet, the outlook for this discipline isn’t too bright. The reason is the very way artificial intelligence works. Made in the image of the human brain, artificial intelligences consist of so-called “neural networks”. A neural network is more or less a – layered – mesh of nodes. The strength of the connections between these nodes determines how the input to the network determines the output. Training the AI means varying the strengths of these connections in a way that the network finally translates the input into a desired output in a decent manner. There are different topologies for these networks, tailored to certain classes of problems but the thing as such is rather universal. Hence AI projects can be rather simple by IT standards: define the right target function, collect proper training data, plug that data to your neural network, train it … . It takes but a couple of weeks and voila, you have an artificial intelligence thatyou can throw on new data for solving your problem. In short, what we can call “intelligence” is the state of strengths of all the connections in your network. The number of these connections can be huge and the nature of the neural network is actually agnostic to the problem you want it to solve. “Unboxing” would thus mean to backwardly extract specific criteria from such a huge and agnostic network. In our radiologist case for example, we would have to find something like “serrated fringes” or “solid core” in nothing but this set of connection strengths in our network. Have fun! Well, you might approach the problem differently by simply probing your AI in order to learn that and how it actually reacts to serrated fringes. But that approach has its limits, too. If you don’t know what to look for or if the results are determined not by a single criterion but by the entirety of some data, looking for specifics becomes utterly difficult. Think of AlphaZero again. It develops strategies and moves that have been unknown to man before. Can we really claim we must understand the logic behind, neglecting the fact that Go as such has been quite resistant to straightforward tactics and logic patterns for the centuries humans have played it. The question is: why “unboxing” after all? – Have you ever asked for unboxing a fellow human’s brain? OK, being able to do that for your adolescent kids’ brains would be a real blessing! But normally we don’t unbox brains. Why are we attracted by one person and not by another? Is it the colour of her eyes, her laughter lines, her voice, her choice of words …? Why do we find one person trustworthy and another one not? Is it the way she stands, her dress, her sincerity, her sense of humour? How do we solve a mathematical problem? Or a business one? When and how do the pieces fall into place? Where does the crucial idea emerge from? Even when we strive to rationalize our decision making, there always remain components we cannot properly “unbox”. If the problem at hand is complex – and thus relevant – enough. We “factor in” strategic considerations, assumptions about the future, others’ expectations … . Parts of our reasoning are shaped by our personal experiences, our individual preferences, like our risk-appetite, values, aspirations, … . Unbox this! Humankind has learnt to cope with the impossibility of “unboxing” brains or lives. We probe others and if we’re happy with the results, we start trusting. We cede responsibilities and continue probing. We cede more responsibilities … and sometimes we are surpassed by the very persons we promoted. Ah, I am entering philosophical grounds again. Apologies! To make it short. I admit, there are some cases in which you might need full transparency, complete “unboxing”. And in case you don’t get it, abolish the idea of using AI for the problem you had in mind. But there are more cases in which the desire for unboxing is just another pretense for not chartering new territory. If it’s intelligent if it behaves like a human why do we ask for so much more from the machines than we would ask from man? Again, I am drifting off into questions of dangerously fundamental nature. Let’s assume for once that we have overcome all our concerns, prejudices and excuses, that despite all of them, we have a business problem we full-heartedly want to throw artificial intelligence at. Then comes the biggest challenge of all. The biggest challenge of all: how to operationalize it Pretty much like in our discussion at the beginning of this post, on the face of it, it looks simple: unplug the human intelligence occupied with the work at hand and plug in the artificial one. If it is significant – quite some AI projects are still more in the toy category – this comes along with all the challenges we are used to in what we call change management. Automating tasks comes with adapting to new processes, jobs becoming redundant, layoffs, re-training and rallying the remaining workforce behind the new ways of working. Yet changes related to artificial intelligence might have a very different quality. They are about “intelligence” after all, aren’t they? They are not about replacing repetitive, sometimes strenuous or boring work like welding metal or consolidating accounting records, they dig to the heart of our pride. Plus, the results are by default neither perfect nor “unboxable”. That makes it very hard to actually operationalize artificial intelligence. Here’s an example. It is more than fifteen years old, taking place at a time when a terabyte was an still an incredible amount of storage, when data was still desired to be stored in warehouses and not floating around in lakes or oceans and when true machine learning was still a purely academic discipline. In short: the good old times. This gives us the privilege to strip the example bare of complexity and buzz. At that time, I was together with a few others responsible for developing Business Intelligence solutions in the area of insurance sales. We had our dispositive data stored in the proverbial warehouse, some smart actuaries had applied multivariate statistics to that data and hurrah, we got propensities to buy and rescind for our customers. Even with the simple means we had by then, these propensities were quite accurate. As an ex-post analysis showed, they hit the mark at 80% applying the relevant metrics. Cutting the ranking at rather ambitious levels, we pushed the information to our agents: customers who with a likelihood of more than 80% were to close a new contract or to cancel one … or both. The latter one sounds a bit odd, but a deeper look showed that these were indeed customers who were intensely looking for a new insurance without a strong loyalty. – If we won them, they would stay with us and loyalty would improve, if a competitor won them, they would gradually transfer their portfolio to him. You would think that would be a treasure trove for any salesforce in the world, wouldn’t you? Far from it! Most agents either ignored the information or – worse – they discredited it. To the latter purpose, they used anecdotal evidence: “My mother in law was on the list”, they broadcast, “she would never cancel her contract”. Well, some analysis showed that she was on the list for a reason but how would you fight a good story with the intricacies of multivariate statistics? Actually, the mother-in-law issue was more of a proxy for a deeper concern. Client relationship is supposed to be the core competency of any salesforce. And now, there comes some algorithm or artificial intelligence that claims to understand at least a (major) part of that core competency as good as that very salesforce … . Definitely a reason to fight back, isn’t it? Besides this, agents did not use the information because they regarded it not too helpful. Many of the customers on the high-propensity-to-buy-list were their “good” customers anyway, those with who they were in regular contact already. They were likely indeed to make another buy but agents reasoned they would have contacted them anyway. So, don’t bother with that list. Regarding the list of customers on the verge of rescinding, the problem was a different one. Agents had only very little (monetary) incentive to prevent these from doing so. There was a recurring commission but asked whether to invest valuable time into just keeping a customer or going for new business, most were inclined to choose the latter option. I could continue on end with stories around that work, but I’d like to share only one more tidbit here before entering a brief review of what went wrong: What was the reaction of management higher up the food-chain when all these facts trickled in? Well, they questioned the quality of the analysis and demanded to include more – today we would say “bigger” – data in order to improve that quality, like buying sociodemographic data which was the fad at that time. Well, that might have increased the quality from 80% to 80+% but remember the discussion we had around redundancy of data. The type of car you drive or the sum covered by your home insurance might say much more than sociodemographic data based on the area you live in. … Not to speak of that eternal management talk that 80% would be good enough. What went wrong? First, the purpose of the action wasn’t thought through well enough from the start. We more or less just choose the easiest way. Certainly, the purpose couldn’t have been to provide agents with a list of leads they already knew were their best customers. From a business perspective the group of “second best customers” might have been much more attractive. Approaching that group and closing new contracts there would have not only created new business but also broadened the base of loyal customers and thus paved the way for longer term success. The price would of course have been that these customers would have been more difficult to win over than the “already good” ones so that agents would have needed an incentive to invest effort into this group. Admittedly going for the second-best group would have come with more difficulties. We might have faced for example many more mother-in-law anecdotes. Second, there was no mechanism in place to foster the use of the information. Whether the agents worked on the leads or not didn’t matter so why should they? Worse even with the churn-list. From a long-term business perspective, it makes all the sense in the world to prevent customer churn as winning new customers is way more expensive. It also makes perfect sense to try making your second-best customers more loyal but from a short-term salesman’s or -woman’s perspective boiling the soup of already good customers makes more short-term sense. Thus, in order to operationalize AI target systems might need a thorough overhaul. If you are serious, that is. The same holds true if you would for example want to establish machine assisted sentiment analysis in your customer care center. Third, there was no good understanding of data and data analytics neither on the supposed-to-be users’ side nor on the management side. This led to the “usual” reflexes on both sides: resistance on the one side and an overly simplified call for “better” on the other one. Whatever “better” was supposed to mean. Of course, neither the example nor the conclusions are exhaustive, but I hope they help illustrate the point: more often than not it is not the analytics part of artificial intelligence that is the tricky one. It is tricky indeed but there are smart and experienced people around to deal with that type of tricky business. More often than not, the truly tricky part is to put AI into operations, to ask the right questions in the first place, to integrate the amazing opportunities in a consistent way into your organization, processes and systems, to manage a change that is more fundamental than simple automation and to resist the reflex that bigger is always better!   So much for today from “Mediocrity and Madness”, the podcast that usually deals with the ever-growing gap between corporate rhetoric and action. I dearly thank all the people who provided inspiration and input to these musings especially in and around the programs I mentioned in the intro, most notably Gemma Garriga, Marcela Schrank Fialova, Christiane Konzelmann, Stephanie Schneider, Arnaud Michelet and the revered Prof. Jürgen Schmidhuber! Thank You for listening … and I hope to have you back soon!  

LA INTELIGENCIA ARTIFICIAL CAMBIARÁ TU VIDATres de cada diez puestos de trabajo podrían desaparecer en 15 años, la existencia de máquinas que piensan por si solas y toman decisiones de manera autónoma ya no está sólo en las películas, de hecho vivimos con ello. Cuando nuestro correo electrónico decide que mensajes son spam y los envía directamente a la papelera, cuando hablamos con Siri en el iPhone o cuando Netflix nos sugiere que series ver, la Inteligencia Artificial entra en juego y casi sin darnos cuenta es algo cotidiano.Se avecina una revolución tecnológica que nos afectará a todos, los expertos creen que será positiva pero algunos trabajos desaparecerán y en otros será necesario que el trabajador se adapte y complete su trabajo con máquinas que aporten esa IA, por ejemplo, pensad en un médico, sin duda seguirá viendo a sus pacientes pero se apoyará en robots que le ayuden en los diagnósticos, operaciones, tratamientos. Otros trabajos como el de traductor, dejará de ser necesario, hoy día hay aplicaciones que traducen instantáneamente. Los trabajos mecánicos y repetitivos los harán robots.Dejaremos de conducir nuestro coche, será más fácil detectar enfermedades, más fácil predecir catástrofes, el consumo, la economía, todo se verá afectado.He leído que en 2016 Microsoft creó un robot dotado de IA que tenía capacidad de aprendizaje, lo presentó en Twitter y cuanto más hablaba más aprendía de los usuarios, otro ejemplo fue Alpha Go Zero, aprendió a jugar al ajedrez y en pocos días ya era capaz de ganar al mejor programa del mundo, ya no hablamos de ganar a un humano, hablamos de ganar a un programa considerado imbatible. Ha habido otros robots que incluso se han llegado a desconectar porque crearon un idioma en Facebook que los humanos no comprendían.Todo esto asusta un poco. Aún no somos muy conscientes de la velocidad a la que la IA se desarrolla, sólo los investigadores y los altos cargos responsables de la IA debaten sobre la ética de algunos métodos, los ciudadanos desconocemos aún las consecuencias de esta tecnología que poco a poco ha entrado en nuestra vida.Canción: El Mundo Futuro. MecanoARTIFICIAL INTELLIGENCE WILL CHANGE YOUR LIFEThree out of ten jobs could disappear in 15 years, the existence of machines that think for themselves and make decisions autonomously is no longer only in the movies, in fact we live with it. When our email decides which messages are spam and sends them directly to the trash, when we talk to Siri on the iPhone or when Netflix suggests which series to watch, Artificial Intelligence comes into play and almost without realizing it is something used daily.A technological revolution, that will affect us all, is approaching, the experts believe that it will be positive but some jobs will disappear and in others it will be necessary for the worker to adapt and complete his work with machines that provide this AI, for example, think of a doctor, No doubt he will continue to see his patients but will rely on robots to help him in the diagnosis, operations, treatments. Other jobs such as translator will no longer be necessary, today there are applications that translate instantly. Mechanical and repetitive work will be done by robots.We will stop driving our car, it will be easier to detect diseases, easier to predict catastrophes, consumption, the economy, everything will be affected.I have read that in 2016 Microsoft created a robot equipped with AI that had learning capacity, presented it on Twitter and the more he spoke the more he learned from users, another example was Alpha Go Zero, he learned to play chess and in a few days it was already able to win the best program in the world, we no longer talk about winning a human, we talk about winning a program considered unbeatable. There have been other robots that have even come to disconnect because they created a language on Facebook that humans did not understand.All this scares a little. We are still not very aware of the speed at which the AI ​​develops, only the researchers and senior officials responsible for AI discuss the ethics of some methods, citizens are still unaware of the consequences of this technology that has gradually entered In our life.

LA INTELIGENCIA ARTIFICIAL CAMBIARÁ TU VIDATres de cada diez puestos de trabajo podrían desaparecer en 15 años, la existencia de máquinas que piensan por si solas y toman decisiones de manera autónoma ya no está sólo en las películas, de hecho vivimos con ello. Cuando nuestro correo electrónico decide que mensajes son spam y los envía directamente a la papelera, cuando hablamos con Siri en el iPhone o cuando Netflix nos sugiere que series ver, la Inteligencia Artificial entra en juego y casi sin darnos cuenta es algo cotidiano.Se avecina una revolución tecnológica que nos afectará a todos, los expertos creen que será positiva pero algunos trabajos desaparecerán y en otros será necesario que el trabajador se adapte y complete su trabajo con máquinas que aporten esa IA, por ejemplo, pensad en un médico, sin duda seguirá viendo a sus pacientes pero se apoyará en robots que le ayuden en los diagnósticos, operaciones, tratamientos. Otros trabajos como el de traductor, dejará de ser necesario, hoy día hay aplicaciones que traducen instantáneamente. Los trabajos mecánicos y repetitivos los harán robots.Dejaremos de conducir nuestro coche, será más fácil detectar enfermedades, más fácil predecir catástrofes, el consumo, la economía, todo se verá afectado.He leído que en 2016 Microsoft creó un robot dotado de IA que tenía capacidad de aprendizaje, lo presentó en Twitter y cuanto más hablaba más aprendía de los usuarios, otro ejemplo fue Alpha Go Zero, aprendió a jugar al ajedrez y en pocos días ya era capaz de ganar al mejor programa del mundo, ya no hablamos de ganar a un humano, hablamos de ganar a un programa considerado imbatible. Ha habido otros robots que incluso se han llegado a desconectar porque crearon un idioma en Facebook que los humanos no comprendían.Todo esto asusta un poco. Aún no somos muy conscientes de la velocidad a la que la IA se desarrolla, sólo los investigadores y los altos cargos responsables de la IA debaten sobre la ética de algunos métodos, los ciudadanos desconocemos aún las consecuencias de esta tecnología que poco a poco ha entrado en nuestra vida.Canción: El Mundo Futuro. MecanoARTIFICIAL INTELLIGENCE WILL CHANGE YOUR LIFEThree out of ten jobs could disappear in 15 years, the existence of machines that think for themselves and make decisions autonomously is no longer only in the movies, in fact we live with it. When our email decides which messages are spam and sends them directly to the trash, when we talk to Siri on the iPhone or when Netflix suggests which series to watch, Artificial Intelligence comes into play and almost without realizing it is something used daily.A technological revolution, that will affect us all, is approaching, the experts believe that it will be positive but some jobs will disappear and in others it will be necessary for the worker to adapt and complete his work with machines that provide this AI, for example, think of a doctor, No doubt he will continue to see his patients but will rely on robots to help him in the diagnosis, operations, treatments. Other jobs such as translator will no longer be necessary, today there are applications that translate instantly. Mechanical and repetitive work will be done by robots.We will stop driving our car, it will be easier to detect diseases, easier to predict catastrophes, consumption, the economy, everything will be affected.I have read that in 2016 Microsoft created a robot equipped with AI that had learning capacity, presented it on Twitter and the more he spoke the more he learned from users, another example was Alpha Go Zero, he learned to play chess and in a few days it was already able to win the best program in the world, we no longer talk about winning a human, we talk about winning a program considered unbeatable. There have been other robots that have even come to disconnect because they created a language on Facebook that humans did not understand.All this scares a little. We are still not very aware of the speed at which the AI ​​develops, only the researchers and senior officials responsible for AI discuss the ethics of some methods, citizens are still unaware of the consequences of this technology that has gradually entered In our life.

自从AlphaGo战胜人类，获得了世界冠军，人工智能就走在了时代的浪潮前。那么世界冠军AlphaGo是无师自通的吗？AlphaGo都有哪些版本？AlphaGo的关键技术是什么？中国科学院计算技术研究所研究员何清，带领我们走进AlphaGo的神奇世界。“SELF格致论道”是中国科学院全力推出、中国科普博览承办的科学讲坛，致力于精英思想的跨界传播。登陆官方网站（self.org.cn）、微信公众号“SELFtalks”获取更多信息。

自从AlphaGo战胜人类，获得了世界冠军，人工智能就走在了时代的浪潮前。那么世界冠军AlphaGo是无师自通的吗？AlphaGo都有哪些版本？AlphaGo的关键技术是什么？中国科学院计算技术研究所研究员何清，带领我们走进AlphaGo的神奇世界。“SELF格致论道”是中国科学院全力推出、中国科普博览承办的科学讲坛，致力于精英思想的跨界传播。登陆官方网站（self.org.cn）、微信公众号“SELFtalks”获取更多信息。

22 – Alpha Zero, Hemimastigotes, Insight en Marte y mares calientes Porque vivimos en un estado de constante fascinación con ella, la inteligencia artificial vuelve a nosotros, en forma de una jugadora que puede aprender a ser la campeona absoluta de todo en más de un juego (perdiste, Alpha Go Zero). También, nos sorprendemos con una re-estructuración considerable al árbol de la vida animal, y platicamos de la llegada de la sonda Insight a Marte, que al mismo tiempo nos sorprende y nos parece normal. Y ya. Después de eso no hay nada más en el archivo de audio, mucho menos un segmento sobre cambio climático claro que no, no seríamos capaces, por favor. Menú 00:09 - Intro 02:54 - Alpha Go Zero aprende a jugar algo más que Go y se convierte en sólo Zero. 09:47 - Los hemimastigotes tienen su propia rama del árbol de la vida 16:55 - Insight llega a Marte 26:57 - ¡Sólo buenas cosas! ¡Yeeei! Mar y calorcito, yeei. Voces y contenido: Mauricio Ortega, Sofía Flores, Rodrigo Pacheco y Víctor Hernández. Edición y producción: Víctor Hernández Voz en la rúbrica: Valeria Sánchez. No se ha mencionado a Elon Musk en todo este año, así que vamos a dejarlo descansar, indefinidamente. Tendría que ganarse un premio nobel o algo así. Este podcast es producido desde un lugar no determinado de la Ciudad de México, donde ya empieza a sentirse frijolito. Música Intro y salida: Little Lily Swing, de Tri-Tachyon, bajo una licencia Creative Commons 3.0 de Atribución: freemusicarchive.org/music/Tri-Tachyon/ Rúbrica: Now son, de Podington Bear, freemusicarchive.org/music/Podington_Bear/ Bajo una licencia Creative Commons Internacional de Atribución No Comercial 3.0 Audios "Daddy's Car: a song composed by Artificial Intelligence - in the style of the Beatles", del canal de youtube Sony CSL. https://www.youtube.com/watch?v=LSHZ_b05W7o Es tal cual, una canción que tiene algunas partes compuestas con inteligencia artificial y algunas con la inteligencia de un compositor francés que se llama Benoît Carré. El sitio del proyecto que la generó es este: http://www.flow-machines.com/ ¿Les parece que sí suena a los Beatles? Fragmento de una entrevista a Alastair Simpson, por CBC News, medio público de Canadá https://www.cbc.ca/news/technology/hemimastigotes-supra-kingdom-1.4715823 Reacción del Laboratorio de Propulsión a Chorro de la NASA cuando amartiza Insight, del canal de youtube SciNews: https://www.youtube.com/watch?v=BiufjRUmndE "Buenas noticias" de Jarabe de Palo (sí, a nosotros también nos sorprendió saber que era de ellos).

欢迎收听丽莎老师讲机器人，想要孩子参加机器人竞赛、创意编程、创客竞赛的辅导，找丽莎老师！欢迎添加微信号：153 5359 2068，或搜索微信公众号：我最爱机器人。丽莎老师讲机器人之最强棋类AI诞生AlphaZero制霸围棋后，又拿下国际象棋和日本将棋。DeepMind最强棋类算法AlphaZero今天以Science封面论文形式发表。经过同行评议，解读这一棋类终极算法，以及实现通用学习系统的重要步骤，正式进入学界和公众的视野。一年前，DeepMind静静地贴出了AlphaZero的预印版论文，当即就在圈内引发轰动：AlphaZero从零开始训练，2小时击败最强将棋AI，4小时击败最强国际象棋AI，8小时击败最强围棋AI AlphaGo ）。现在，DeepMind将完整评估后的AlphaZero公之于众，不仅验证了上述结果，还补充了新的提升。AlphaZero没有使用人类知识（除了棋类基本规则），从零开始训练，快速掌握日本将棋、国际象棋和围棋这三种复杂棋类游戏，展现出令人耳目一新的独道风格，拓展了人类智慧，并证明了机器拥有创造性的可能。20年前输给IBM深蓝的国际象棋世界冠军卡斯帕罗夫，今天在Science发表社论，表示他很高兴看到AlphaZero展现出了像他一样“动态、开放”的棋风：“传统观点以为，机器将通过无休止的枯燥操作趋近完美，最终导致平局。但据观察，AlphaZero优先考虑棋子的活动而非盘面上的点数优势，更喜欢在我看来有风险和激进的地方落子。“计算机程序通常会反映出编程者的侧重和偏见，但由于AlphaZero通过自我对弈训练，我认为它体现了棋的真谛（truth）。正是这种出色的理解使其能够超越世界顶级的传统棋类引擎，而且每秒计算的落子位置要少得多。”AlphaZero证明了机器也能成为专家，机器生成的知识也值得人类去学习。“AlphaZero以这样一种强大而有用的方式超越了我们，只要在虚拟知识（virtual knowledge）能够生成的领域，这个模型都可能复制到任何其他任务上。”DeepMind论文使用通用的搜索方法，结合蒙特卡罗树搜索（MCTS），增强了深度强化学习。“尽管MCTS已经成为围棋程序中的标准搜索方法，但迄今为止，几乎没有证据表明它在国际象棋或将棋中有用。”Campbell写道：“DeepMind展示了深度强化学习与MCTS算法相结合的力量，从随机初始化的参数开始，让神经网络通过自我对弈不断更新参数。”下面，就让我们一起来看看，AlphaZero的几位论文作者阐述他们如何用5000个TPU，让AlphaZero快速掌握将棋、国际象棋和围棋。传统国际象棋的引擎依赖于由人类高手玩家“手工制作”的数千条规则和启发式方法，它们都试图解释游戏中可能发生的每一种结果。日本将棋程序也是特定于游戏的，使用与国际象棋程序类似的搜索引擎和算法。AlphaZero则采用了一种完全不同的方法，用深度神经网络和通用算法取代了这些“手工制作”的规则，而这些算法对基本规则之外的游戏却一无所知。在国际象棋中，AlphaZero仅用了4个小时便首次超越了Stockfish；在日本将棋中，AlphaZero在2小时后首次超过Elmo；在围棋方面，AlphaZero在2016年的比赛中，经过30个小时的鏖战，首次击败了传奇棋手李世石。注：每个训练步骤代表了4096个落子位置。为了学习每一个游戏，一个未经训练的神经网络通过强化学习与自己对打数百万次。一开始，它完全是随机的，但是随着时间的推移，系统从输赢中开始学习，并根据神经网络的参数进行调整，使其在未来可以选择更有利的走法。网络需要的训练量取决于游戏的风格和复杂性，国际象棋需要9小时，将棋需要12小时，围棋需要13天。AlphaZero的一些举动，例如将王将移至棋盘中央是有违将棋理论的，从人类的角度来看，它的这些举动似乎是将自己置于危险境地。但令人难以置信的是，它仍然控制着局面。AlphaZero独特的游戏风格向我们展示了将棋的新可能性。训练后的网络用于指导搜索算法（蒙特卡罗树搜索，MCTS），选择游戏中最有有利的动作。对于每次移动，AlphaZero仅搜索传统国际象棋引擎所考虑的一小部分位置。例如，在国际象棋中，它每秒仅搜索6万个位置，相比之下，Stockfish大约有6千万个位置。这些经过全面训练的系统是在国际象棋AI Stockfish和将棋AI Elmo最强大的“手工引擎”以及我们之前自学的AlphaGo Zero系统(已知最强大的围棋选手)的帮助下进行测试的。每个程序都在它们所设计的硬件上运行。Stockfish和Elmo使用了44个CPU核，而AlphaZero和AlphaGo Zero使用了一台拥有4个第一代TPU和44个CPU核的机器。第一代TPU在推理速度上与NVIDIA Titan V GPU等商用硬件大致相似，但架构并不具有直接可比性。所有的比赛都有时间控制，每场比赛3小时，外加每一步额外的15秒。在每次评估中，AlphaZero都毫无悬念地击败了对手：在国际象棋比赛中，AlphaZero击败了2016年TCEC(第九季)世界冠军Stockfish，赢得155场比赛，在1000场比赛中只输了6场。为了验证AlphaZero的稳健性，我们还进行了一系列比赛，这些比赛都是从常见的“人类开局方式”开始的。在每一种开局情况下，AlphaZero都击败了Stockfish。我们还与最新开发版本的Stockfish以及它的变体打过比赛，在所有的比赛中，AlphaZero都赢了。在将棋比赛中，AlphaZero击败了2017年CSA世界冠军版Elmo，赢得了91.2％的比赛。在围棋比赛中，AlphaZero击败了AlphaGo Zero，赢得了61％的比赛。独创棋风，拓展人类智慧，迈向通用学习系统重要一步让人最着迷的是AlphaZero的行棋风格。例如，在国际象棋中，AlphaZero在自我训练中独立发现并走出了人类棋手常用的定式，如开局、王不立险地（King safety）和兵的走法。但是，由于这些都是自学的，因此不会受传统观念的影响，AlphaZero还开创出了自己的直觉和策略，产生了一系列令人兴奋的新颖思路，为几个世纪以来国际象棋战略战术的思考提供了有益的补充。过去一个多世纪以来，国际象棋一直被用作衡量人类和机器认知水平的黄金标准。 AlphaZero取得的非凡成果，刷新了这门古老的棋盘游戏和尖端科学之间的显著联系。在与AlphaZero对弈时，棋手注意到的第一件事就是它的行棋风格，AlphaZero的行棋风格非常灵活，最大限度地提升己方子力配备的灵活性和机动性，同时最大限度地降低对手子力的灵活性和机动性。与我们的通常想法不同的是，AlphaZero似乎对“子力”本身的重视程度较低，而重视“子力”是现代国际象棋的基本行棋思路，棋盘上每个子都具有价值，如果一个玩家在棋盘上的子力高于对手，那么他就拥有子力优势。而AlphaZero甚至愿意在棋局早期牺牲子力，以获得长期收益。令人印象深刻的是，AlphaZero在行棋时能将这种风格应用在各种各样的开局和定式中。AlphaZero从走第一步开始就体现出了这种明确的的性，且一以贯之，其风格体现得非常明显。过去的传统国际象棋软件已经非常稳定，几乎不会出现明显错误，但在面对没有具体和可计算解决方案的时，其行棋会发生偏差，正是在这种时候，才是AlphaZero发挥其'感觉'、'洞察'或'直觉'的地方。这种独特的能力，在其他传统的国际象棋引擎中是看不到的。目前，AlphaZero已经被用来在世界国际象棋锦标赛上为棋迷们提供有关Magnus Carlsen和Fabiano Caruana（现男子国际象棋等级分前两名）对局的新见解和评论。我们可以看看AlphaZero的分析，与顶级国际象棋大师对棋局的分析，甚至和棋手实战着法有何不同，这真是令人着迷的一件事。AlphaZero可以作为整个国际象棋社区的强大教学工具。AlphaZero的“教诲”，让我们想起了2016年AlphaGo与围棋世界冠军李世乭对弈时的场景。在那次比赛中，AlphaGo走出了许多极具创造性的致胜着法，包括在第2局比赛中的执黑第37手，这手棋推翻了人类数百年的思路。这些着法已经被包括李世乭本人在内的所有级别的棋手和爱好者研究过。之前还认为AlphaGo是基于概率来计算的，它只是一台机器。但当棋手看到这手棋时，改变了想法。毫无疑问，AlphaGo是有创造性的。“人机大战”的影响力已经远远超出了国际象棋本身。这些自学成才的专家级机器不仅表现优异，棋力非凡，而且从自己创造的新知识中学习。和围棋一样，我们对AlphaZero在国际象棋上的创造性突破感到兴奋，自从计算机时代以来，人工智能时时面临着巨大挑战，包括巴贝奇、图灵、冯·诺依曼在内的早期计算机先驱人物，都曾试图设计国际象棋程序，但AlphaZero的用途不仅仅是国际象棋、将棋和围棋。为了创建能够解决各种现实问题的智能系统，它们需要更加灵活，能够适应新情况。虽然目前在实现这一目标方面取得了一些进展，但AI的通用化问题仍然是研究中的一项重大挑战，经过训练的AI系统面对特定任务时能够以极高标准完成，但任务只要稍有变化往往就会失败。AlphaZero掌握了三种不同的复杂游戏，这可能是朝着解决这一问题迈出的重要一步。尽管目前还处于早期阶段，但AlphaZero取得的进步，以及在蛋白质折叠系统AlphaFold等其他项目上的令人鼓舞的结果，让我们对实现通用学习系统的使命充满信心，相信未来我们能够找到一些新的解决方案，解决最重要、最复杂的科学问题。

Worden machines slimmer dan mensen? Bennie Mols, die voor NRC over artificiële intelligentie schrijft, denkt van niet. Natuurlijk: ze kunnen beter schaken dan de beste schaker en verslaan de wereldkampioen in Go. Maar het kost veel energie, en het inlevingsvermogen van een peuter is nog altijd groter. Hoe intelligent is kunstmatige intelligentie wel?Presentatie: Lucas Brouwers en Hendrik SpieringProductie: Mirjam van Zuidam@lucasbrouwers // @hendrikspiering // @BennieMolsKijk hier de video over het inlevingsvermogen van peuters: https://www.youtube.com/watch?v=Z-eU5xZW7cUOf check hier de afleveringen van Dokters vs Internet: https://www.npostart.nl/dokters-vs-internet/03-05-2018/KN_1697746Lees ook het artikel dat Bennie Mols over peuters en robots schreef: https://www.nrc.nl/nieuws/2018/11/09/waarom-computers-het-afleggen-tegen-peuterintuitie-a2754648Of luister naar de stem van de eerste robot van de toekomst HAL9000 uit A Space Oddyssey: https://www.youtube.com/watch?v=QFSE4dUJYM8

“Data is the new oil” has become a popular declaration in headlines circulating around Silicon Valley, but in this episode, we question the veracity of the phrase. The argument for equating data to oil is that data will be the resource that will shape the 21st century in the way that oil shaped the previous century. While data, like oil, needs to be refined in order to be useful, it's not necessarily true that the more data you have, the more of a competitive advantage you have. Or… is it?   Show notes   Data is the new oil… or is it? (0:57) Not all data is created equal (3:31) The All Turtles article about Moorfields Hospital in London's use of data from eye scans (3:38) AlphaGo Zero: learning from scratch (DeepMind) (4:11) The distinction between public data and private data (8:55) Kaggle has tens of thousands of datasets (9:10) Who should be able to profit from your data? (13:16)   “Eyeroll, please.” Debunking the common startup advice to “start local.” (19:22) Avoiding building a product that only serves a bubble. The problem with thinking of countries as markets.   Listener question (24:45) From Ari via email: I'd like to hear your podcasting team's reaction to and solution for the issue of algorithmic learning beyond the control of app developers. Leave us a voicemail with your question and we'll play it on a future episode: +1 (310) 571-8448 (29:48)   We want to hear from you Please send us your comments, suggested topics, and listener questions for future All Turtles Podcast episodes. Voicemail: +1 (310) 571-8448 Email: hello@all-turtles.com Twitter: @allturtlesco with hashtag #askAT For more from All Turtles, follow us on Twitter, and subscribe to our newsletter on our website.

This is a MarketScale Software & Technology Podcast Series, hosted by Daniel Litwin. This is the first episode of a three part series titled Bringing AI to Businesses with Ben Taylor, Chief AI Officer & Co-founder of ZIFF Inc. In each episode, we'll explore different aspects of AI's push into business-operations ubiquity, from its most useful applications to the surprising business ethics that come with implementation. Each episode will also feature a short article from Taylor, which you can read below. How Do You Deliver on an AI Project, Both As an Executive and a Data Scientist? Many executives are naive when it comes to AI capabilities and navigating where AI might provide value to their business. The data scientists aren't helping either, where most struggle to communicate value to the business representatives. Most data scientists also lack urgency; they have no pressure to last-mile AI into production. Funding science projects will accomplish one thing: covering the tuition of your data science team so they can land better jobs at Facebook or Google. So, once the trigger is pulled and you have a team prepped and passionate about bringing AI to your business...how do you make sure everyone delivers so you avoid wasted time and money? Avoiding The Science Project Landmines If you are an internal advocate for AI, do everything you can to constrain the timeline. Ask yourself: Is there anyway to do an internal proof of concept in 60 days instead of 12 months? What can I do to reduce internal budget? What can I do to reduce the number of people required? The more you reduce these variables, the more likely you are to get buy-in from the internal business units. I've always been a fan of leveraging outside hardware companies, consulting groups, or AI platforms to shorten do-ability tests. Crawl, Walk, Run Some AI projects fail because they are too ambitious. They don't have a short-term proof-point, and the complexity comes tumbling down like a house of cards, revealing a project that had no clear goals, tangible value or structure. This flaw can come from inexperienced data science teams that are too "academic," where they are more interested in a challenging thought experiment than a Bayesian method in production. If you can carve your project up into bite size milestones, your chances of success are higher. It shouldn't be ignored that AI projects aren't a one-and-done either; you have the advantage of improving on your algorithms. Just look at the evolution between AlphaGo and AlphaGo Zero, and how they would've never achieved such a grand level of "reinforcement learning" without trying a few, more tangible methods first. Get some novice wins into production and then level up on subsequent versions. It Is Harder Than It Looks Getting a successful AI project to value is much harder than it looks. Most major wins for AI are behind six to 10 iterations on the same problem. We see successful companies solving the same problem multiple times, where each time they solve it they understand the data set and problem a little better. Once a project has crossed a predefined criteria for success, taking that AI project into production can create additional problems. Supporting AI in production requires quality monitoring (e.g. did your incoming features drift) to ensure models are behaving as designed. This requires an involved data science team. And yes, I said team. Collaborate, get multiple eyes on the project, and make sure everyone is on the same page before launching something into production. You don't want your AI project to end up like the Mars Climate Orbiter: dead in the air because of a unit conversion mistake. Double check, triple check, and then check again that the final product is in line with the initial vision you set up for success. Feels a lot like simple project management, huh? Highlights from the Episode [transcription] For the latest news, videos, and podcasts in the Software & Electronics Industry, be sure to subscribe to our industry publication. Follow us on social media for the latest updates

Jason and Kilian discuss artificial intelligence (AI) and its implications for our future. If you've ever heard a discussion of AI and felt like an idiot because you couldn't understand what the scientists were talking about, then this conversation is for you. We sound like idiots and don't get any of the science, but we're scared anyway. AlphaGo Zero: https://deepmind.com/blog/alphago-zero-learning-scratch/ DeepMind: https://deepmind.com/ Sam Harris Ted Talk: https://www.ted.com/talks/sam_harris_can_we_build_ai_without_losing_control_over_it Sam Harris Podcast with Eliezer Yudkowsky: https://samharris.org/podcasts/116-ai-racing-toward-brink/ Telegraph article: https://www.telegraph.co.uk/news/2018/05/28/robots-will-never-rise-against-humans-part-family-ai-expert/

Mit neuem technischen Setup ging es diesmal vor allem um Intelligenz. Zu Beginn gab es eine kurze Ergänzung zu Folge 31 und der Frage, warum Diagnosen von Persönlichkeitsstörungen (wie Borderline) bei Kindern und Jugendlichen von Fachleuten als nicht seriös angesehen werden. Das psychologische Einstiegsthema waren Lernstörungen wie Lese- und Rechtschreibstörung, Dyslexie, Dyskalkulie und andere Entwicklungsstörungen schulischer Fertigkeiten. Sebastian berichtete aus seiner gutachterlichen Praxis (v.a. im Zusammenhang mit §35a SGB VIII, Eingliederungshilfe für seelisch behinderte Kinder und Jugendliche). Schnell ging es aber auch um die Abgrenzung von Lernstörungen zu Intelligenzminderungen und damit Themen wie Intelligenzmessung, IQ-Tests und dem Verein Mensa. Anschließend wurde auf Anlaufstellen für von Lernstörungen Betroffene verwiesen, wie Schulpsychologen und Versorgungsämter, und schließlich noch mit dem Mythos der vermeintlichen Lerntypen aufgeräumt. Im Hauptteil ging es dann mit Gast Dr. Dennis Eckmeier um künstliche Intelligenz (KI). Dennis ist promovierter Biologe und arbeitet derzeit als Neurowissenschaftler in Lissabon unter anderem am maschinellen Sehen der Pfotenbewegungen von Mäusen. Was ist maschinelles Lernen und was ein neuronales Netz? Wie unterscheiden sich überwachtes und unüberwachtes Lernen? Und warum ist das alles eher wie WOPR aus dem Film "WarGames" als die Superintelligenz aus anderen Science-Fiction-Szenarien? Haben Algorithmen eine Motivation? Ist bestärkendes Lernen nichts anderes als Evolution? Und müssen Systeme wie AlphaGo Zero zwangsläufig besser werden als Menschen - zumindest wenn es darum geht, Go zu spielen? Alexander verwies dabei auf einen aktuellen Artikel der Universität Stanford, und Sebastian auf den regelmäßigen KI-Report der gleichen Institution. Natürlich wurde ausführlich über Alan Turing und den nach ihm benannten Test gesprochen. Turings bahnbrechende Veröffentlichung von 1950 birgt aber auch noch andere Überraschungen bis hin zum Hellsehen. Alan Turings persönliches "Imitation Game" brachte das Team über das Apple-Logo zum Film "Ex Machina". Kann ein System wie ELIZA einen Therapeuten ersetzen? Wem gehören die Urheberrechte an einem Musikstück, bei der eine KI die Beatles imitiert? Macht es Sinn, zukünftig Friseurtermine mit dem Google Assistant zu vereinbaren? Und sollen wir einer Software wie der Babylon Health App vertrauen? Zum Abschluss wurde es metaphysischer, mit Gesprächen zur Singularität (googeln!), "Solaris" von Stanislaw Lem (Buch lesen!), "Origin" vom Dan Brown (Finger weg!), "Blade Runner" nach Philip K. Dick (Film schauen!), den Podcast "Science for progress" von Dennis (anhören!) und aktuelle Beweise fehlender menschlicher Intelligenz wie bei den Incels (nicht googeln!). Der Titel dieser Folge wurde von Hörer Roland Schuler auf Twitter vorgeschlagen, und zwar bereits Tage vor der Sendung - das ist Intelligenz (oder Hellseherei)!

In this episode I’m joined by legendary chess champion, author, and fellow at the Oxford Martin School, Garry Kasparov. Garry and I sat down after his keynote at the Figure Eight Train AI conference in San Francisco last week. Garry and I discuss his bouts with the chess-playing computer Deep Blue–which became the first computer system to defeat a reigning world champion in their 1997 rematch–and how that experience has helped shaped his thinking on artificially intelligent systems. We explore his perspective on the evolution of AI, the ways in which chess and Deep Blue differ from Go and Alpha Go, and the significance of DeepMind’s Alpha Go Zero. We also talk through his views on the relationship between humans and machines, and how he expects it to change over time. The notes for this show can be found at twimlai.com/talk/140. For more information on this series, visit twimlai.com/trainai2018.

In 2017 gebeurde er iets heel speciaals op het gebied van Artificial Intelligence: het programma AlphaGo Zero wist het programma AlphaGo te verslaan met schaken. AlphaGo Zero had het spel geleerd door tegen zichzelf te spelen. Ben en Herbert vragen daarom in deze Technoloog AI-hoogleraar Aske Plaat het hemd van het lijf.

Fredrik discusses VR with Noah Falstein of the Inspiracy (and previously companies such as Google and Lucasfilm games). We talk about where VR is today, which platforms are good today and what might happen going forward. VR might be on the verge of a big breakthrough but there is still a lot left to be discovered, from ways of controlling experiences to entire new genres. Recorded on stage at Øredev 2017. Thank you Cloudnet for sponsoring our VPS! Comments, questions or tips? We are @kodsnack, @tobiashieta, @oferlund and @bjoreman on Twitter, have a page on Facebook and can be emailed at info@kodsnack.se if you want to write longer. We read everything we receive. If you enjoy Kodsnack we would love a review in iTunes! Links Noah Falstein Lucasfilm games Indiana Jones and the last crusade Noah’s keynote - “The real, the virtual, and the cortex” Noah’s second presentation - Lucasfilm games and the rise of Lucasarts Habitat The Habitat promotional video Club Caribe Quantum link The QWERTY keyboard The OS X dock Google Spotlight stories Pearl Special delivery - by Aardman animation The Simpsons VR episode A nice (360) flight over Pyonyang James Cameron’s Avatar sequels Games for health 40 predictions for VR/AR through 2025 Dataglove Jaron Lanier Apple Newton Polybius Jeff Minter Virtual virtual reality Portal The lost bear Steve Meretzky Planetfall The Sims Doom Castle Wolfenstein Katamari damacy Memory palaces Alphago and Alphago zero The holodeck Dream park, by Larry Niven and Steven Barnes Ready player one Black mirror Titles Nobdy had ever experienced that I become a character in the computer? A realtime, constant back and forth A version that doesn’t allow you to do most of the fun stuff It demos well Still looking for the killer apps in VR The grammar of VR storytelling The Spielberg or Lucas of VR An “of course” moment Something came along and ate the flower I’m tired of watching things eat eachother The Pixar movies of 2020 Hard plastic is actually preferable As scary as they need to be A robot named Floyd We were discovering entirely new genres Put that house into VR Page number 27: things you find in a kitchen

Robots taking jobs. The AI apocalypse. Universal basic income. These themes encompass the fears of many parents as they think about a future for their children. In this week's episode, Phil Libin, Jessica Collier, and Blaise Zerega address this worry and include advice from the likes of Senator Mark Warner, Kai-fu Lee, and Stephen Wolfram. Your hosts also assess notable developments in practical  AI from 2017 and share their daily encounters with iMessage, LinkedIn, and financial services app Penny. Show notes Developments in practical AI from 2017 (00:45) AlphaGo Zero (01:51) Voice assistants go mainstream (04:57) Everyday AI use cases iMessage autofill for calendar (6:12) LinkedIn Premium for U.S. military veterans; Thank you for your service (7:30) Penny, a financial services bot (8:46) Listener question What should I tell my two-year old daughter to help her survive the AI apocalypse? (11:45) Senator Mark Warner, interviewed by Blaise Zerega on May 30, 2017, at the New Deal Ideas Summit in San Francisco. Kai-Fu Lee, interviewed by Blaise Zerega on May 22, 2017, over lunch in San Francisco. Stephen Wolfram, interviewed by Blaise Zerega on May 4, 2017, onstage with Group M's Irwin Gottlieb at the Collision Conference in New Orleans. Cafe X first Robotic Cafe in USA Robot Baristas serve up coffee Meet SAM, the bricklaying robot Building Tomorrow - Robotics in Construction Tomorrow Daily - Japanese construction firm using robotic bulldozers guided by drones, Ep. 257 Wolfram Alpha: What jobs will still be important after the AI revolution? Will Robots Take Our Children's Jobs? (The New York Times) Sapiens: A Brief History of Humankind, by Yuval Noah Harari (2014) Homo Deus: A Brief History of Tomorrow, by Yuval Noah Harari (2016) We want to hear from you Please send us your comments, suggested topics, and questions for future episodes: Email: hello@all-turtles.com Twitter: @allturtlesco with hashtag #askAT For more from All Turtles, follow us on Twitter, and subscribe to our newsletter on our website. Thanks for listening  

El tema principal del episodio son los Bitcoin, desde qué ha pasado con ellos, hasta los trajes para minar monedas virtuales con calor humano, sin olvidarnos de nuestras #predicciones. También hablamos de la nueva pestaña de comunidad de Youtube, la compra de Shazam por parte de Apple, las novedades de AlphaGo Zero, la inteligencia artificial experta en juegos de mesa. Antes de acabar, nuestras recomendaciones (series, gadgets, apps...) que seguro que querrás conocer.

Dernier Geeksleague de la saison 7, Geeksleague reviendra avec une nouvelle saison le 19 Janvier 2018. Pour s'abonner à notre flux RSS Podcast Pour nous soutenir via Tipeee Pour nous écouter sur Dezzer Au sommaire : Les news Tech de la semaine Google Home Pit People Xenoblade Chronicles 2 Quand la technologie menace l’humanité : Capacity crunch et les robots dans la finance "World Community Grid" et le calcul distribué AlphaGo Zero : intelligence artificiel Dragon Quizz Point Remerciements : À nos tipeurs !! Pour en savoir plus : La chaîne Youtube de Renald où il présente des applications au Google Home Dragon Quizz Point : Nicolas Kanaeleff remporte un magnifique Pull Geeksleague  Wally gagne le ramasse miette doré !!! En vidéo :

How do you think the progression of artificial intelligence will affect your career plans and the projects you are working on? If you enjoyed this episode, check out show notes, resources, and more at https://www.superdatascience.com/110

Discutirem per saber qu

Semana 15 - Skynet aún no; una nueva era en la astronomía ya / Parece que la nueva era en la astronomía ha llegado, de la mano del evento astronómico GW17082017 (para más señas, dos estrellas de neutrones en colisión). Víctor y Pach comentan esa historia, adicionalmente a un reporte geológico sobre la cuenca del Valle de México, una región donde los terremotos de septiembre en México dejaron pérdidas de vidas y muchos daños: ese reporte podría ayudar a explicar el nivel del riesgo. Finalmente, Axel Becerril visita el podcast de nuevo, esta vez para hablar largo y tendido de inteligencias artificiales, y explicarnos por qué Skynet aún está lejos y podría terminar siendo una brindadora de conocimiento. ¡Acompáñennos! Menú 00:24 – Intro 01:31 – Detección de estrellas de neutrones en colisión con ondas gravitacionales y luz 13:13 – Un mapa de las fallas geológicas en la Ciudad de México 23:05 – ¿Qué tan lejos está Alpha Go Zero de Skynet? 01:03:38 - Despedida y métodos de contacto Créditos musicales completos en nuestra página de Soundcloud. Este podcast es producido desde un lugar no determinado de México, y a veces en un día determinado que terminan siendo dos fechas por problemas técnicos. Número de veces que se menciona a Elon Musk en este episodio: 1. Ligas de interés Más sobre el descubrimiento de las ondas gravitacionales: http://www.bbc.com/mundo/noticias-41637747 Sobre el reporte geológico de la Ciudad de México: http://www.gaceta.unam.mx/20171016/mapea-la-unam-fracturas-en-el-suelo-de-cdmx/ http://www.geociencias.unam.mx/geociencias/desarrollo/fracturas_sismo2017.pdf Sobre Alpha Go Zero: https://www.unocero.com/noticias/ciencia/alphago-zero-un-avance-de-la-inteligencia-artificial-sin-precedente/ https://www.entrepreneur.com/article/303444 El artículo en la revista Nature (en inglés): https://www.nature.com/articles/nature24270.epdf?author_access_token=VJXbVjaSHxFoctQQ4p2k4tRgN0jAjWel9jnR3ZoTv0PVW4gB86EEpGqTRDtpIz-2rmo8-KG06gqVobU5NSCFeHILHcVFUeMsbvwS-lxjqQGg98faovwjxeTUgZAUMnRQ Y el soundtrack de Blade Runner, porque es una joya: https://www.youtube.com/watch?v=k3fz6CC45ok Música y audios Intro y salida: Little Lily Swing, de Tri-Tachyon, bajo una licencia Creative Commons 3.0 de Atribución: http://freemusicarchive.org/music/Tri-Tachyon/ Noticias comentadas: Instar (Instrumental) por Robin Allender que está licenciada bajo una Licencia Attribution-NonCommercial 3.0 International. Tema discutido: Passage of Time (Duet) por Martijn de Boer (NiGiD) (c) 2016 Licenciada bajo una licencia Creative Commons Attribution Noncommercial (3.0): http://dig.ccmixter.org/files/NiGiD/52856 Ft: Doxent Zsigmond Rúbrica: Now son, de Podington Bear, http://freemusicarchive.org/music/Podington_Bear/ Bajo una licencia Creative Commons Internacional de Atribución No Comercial 3.0 Voz en la rúbrica: Valeria Sánchez. Audio 1: Anuncio de la señal de ondas gravitacionales GW17082017, desde el detector VIRGO https://www.youtube.com/watch?v=mtLPKYl4AHs Audio 2: Espectrograma de la señal, desde el canal de LIGO-Virgo: https://www.youtube.com/watch?v=aWX-BY-A9CY Audio 3: Discurso de John Connor en contra Skynet: https://www.youtube.com/watch?v=pS9MNvJZrdk Voz en la rúbrica: Valeria Sánchez.

在 AlphaGo Zero 以100:0秒杀了它的前辈以后，为了挽回人类最后一丝颜面，我们决定以本期节目再一次挑战人工智能。本期话题我们会围绕人工智能将如何帮助提高人类创造力以及设计师未来的发展方向展开讨论。--------------------------------------------------------------------- 本周四微信群内讨论三个问题： 1. 你认为相对于人类，人工智能的缺点和劣势是什么？有什么是人工智能做不到的呢？2. 节目里提到的未来设计师角色转换会如何影响你的专业方向选择以及自己的职业规划呢？你会转行吗？哈哈哈3. 聊聊你觉得哪些科幻电影里的桥段会在我们有生之年成真呢？当梦想照进现实，生活会变得更美好还是反而更可怕了呢？--------------------------------------------------------------------- 很多朋友问怎么进入神秘的“异能微信群”，请关注我们的公众号“异能FM”，导航栏处点击“加入我们”后点击“加微信群”，点开推送，填写进群问卷后，我们神秘的异能君会联系你的。 商业合作请在电台公众号“异能FM”后台留下您的联系方式，或联系邮件：info@yineng.fm 本期主播：Sai, Oba, Monkey Rider节目剪辑/文：ObaPoster：Sai（Poster图片来自网络）

在 AlphaGo Zero 以100:0秒杀了它的前辈以后，为了挽回人类最后一丝颜面，我们决定以本期节目再一次挑战人工智能。本期话题我们会围绕人工智能将如何帮助提高人类创造力以及设计师未来的发展方向展开讨论。--------------------------------------------------------------------- 本周四微信群内讨论三个问题： 1. 你认为相对于人类，人工智能的缺点和劣势是什么？有什么是人工智能做不到的呢？2. 节目里提到的未来设计师角色转换会如何影响你的专业方向选择以及自己的职业规划呢？你会转行吗？哈哈哈3. 聊聊你觉得哪些科幻电影里的桥段会在我们有生之年成真呢？当梦想照进现实，生活会变得更美好还是反而更可怕了呢？--------------------------------------------------------------------- 很多朋友问怎么进入神秘的“异能微信群”，请关注我们的公众号“异能FM”，导航栏处点击“加入我们”后点击“加微信群”，点开推送，填写进群问卷后，我们神秘的异能君会联系你的。 商业合作请在电台公众号“异能FM”后台留下您的联系方式，或联系邮件：info@yineng.fm 本期主播：Sai, Oba, Monkey Rider节目剪辑/文：ObaPoster：Sai（Poster图片来自网络）

In 2016, the world champion Lee Sedol was beaten at the ancient boardgame of Go - by a machine. It was part of the AlphaGo programme, which is a series of artificially intelligent systems designed by London-based company DeepMind. AlphaGo Zero, the latest iteration of the programme, can learn to excel at the boardgame of Go without any help from humans.So what applications could AI learning independently have for our day-to-day lives? Katie Haylor spoke to computer scientist Satinder Singh from the University of Michigan, who specialises in an area within artificial intelligence called... Like this podcast? Please help us by supporting the Naked Scientists

In 2016, the world champion Lee Sedol was beaten at the ancient boardgame of Go - by a machine. It was part of the AlphaGo programme, which is a series of artificially intelligent systems designed by London-based company DeepMind. AlphaGo Zero, the latest iteration of the programme, can learn to excel at the boardgame of Go without any help from humans.So what applications could AI learning independently have for our day-to-day lives? Katie Haylor spoke to computer scientist Satinder Singh from the University of Michigan, who specialises in an area within artificial intelligence called... Like this podcast? Please help us by supporting the Naked Scientists

V dnešnom podcaste si povieme o zaujímavom prípade z Austrálie o wellness bloggerke Belle Gibson, ktorá si popularitu vybudovala na tom, že tvrdila, že sa z rakoviny mozgu vyliečila len stravou. Ďalšou témou bude umelá inteligencia od Google - AlphaGo Zero, ktorá je, samozrejme, ešte lepšia, ako tá predchádzajúca. TémyYouTubeZdroje Intro Wellness bloggerka Belle Gibson dostala pokutu za podvody AlphaGo Zero Outro https://youtu.be/jPDSo6jwd70 Belle Gibson: Blogger who claimed she beat cancer with 'healthy eating' fined more than $400,000 over lies Health Blogger Gibson Fined Belle Gibson AlphaGo Zero: Learning from scratch

AlphaGo the AI developed to play the ancient board game, Go, crushed 18-time world champion Lee Sedol and the reigning world number one player, Ke Jie. But now, an even more superior competitor is in town. AlphaGo Zero has beaten AlphaGo 100-0 after training for just a fraction of the time AlphaGo needed, and it didn't learn from observing humans playing against each other – unlike AlphaGo. Anthony and Jeff discuss how it did it, and what it means for the future of AI. GET BONUS EPISODES, VIDEO HANGOUTS AND MORE. VISIT: http://patreon.com/wehaveconcerns Get all your sweet We Have Concerns merch by swinging by http://wehaveconcerns.com/shop Hey! If you’re enjoying the show, please take a moment to rate/review it on whatever service you use to listen. Here’s the iTunes link: http://bit.ly/wehaveconcerns And here’s the Stitcher link: http://bit.ly/stitcherwhconcerns Or, you can send us mail! Our address: We Have Concerns c/o WORLD CRIME LEAGUE 1920 Hillhurst Ave #425 Los Angeles, CA 90027-2706 Jeff on Twitter: http://twitter.com/jeffcannata Anthony on Twitter: http://twitter.com/acarboni Today’s story: https://www.inc.com/lisa-calhoun/google-artificial-intelligence-alpha-go-zero-just-pressed-reset-on-how-we-learn.html If you’ve seen a story you think belongs on the show, send it to wehaveconcernsshow@gmail.com, post in on our Facebook Group https://www.facebook.com/groups/WeHaveConcerns/ or leave it on the subreddit:http://reddit.com/r/wehaveconcerns

Kill all Humans Le prochain tube de l’été sera… Google reconnaît Médor. Adobe colorise les portraits, les croquis, et nettoie les vidéos automatiquement. DeepL, un traducteur allemand qui pourrait bien détrôner Google via Korben. AlphaGo Zero apprend tout seul à progresser au jeu de Go. Une Girafe qui joue aux échecs. AutoML: les IA font des bébés ? Probablement pas... Deus Ex Machina ? Elon Musk est pas d’accord... Mais Poutine est plus inquiet des super-soldats génétiquement modifiés... Influences réseaux sociaux et Fake news Sur le Brexit aussi ? Pokémon Go aurait servi les influences russes sur les élections américaines Une solution basée sur la blockchain ? Ou pas ? Les experts sont très partagés. N’oublions pas nos biais ! En tout cas, des ex-salariés continuent de s’alarmer sur ce que devient leur bébé. Twitter a mis un an à désactiver un compte qui posait de furieux soucis. Chiffrement et sécurité Google est bien content de l’évolution du chiffrement avec https Mais le directeur du FBI aimerait pouvoir déchiffrer des smartphones qui font l’objet d’enquêtes. Bruxelles est contre les portes dérobées mais encourage à partager les moyens de déchiffrement. Sécurité : des initiatives pour protéger du minage intempestif… Parce que ça va pas s’arranger... Tiens, un nouveau Malware : Bad Rabbit Au fait, les montres connectées pour enfants sont des passoires... En bref... Le stream-ripping : c’est bien ou pas ? Les animojis, les suites de l’affaire de la marque déposée. Bug de la calculatrice d’iOS 11, une explication du mauvais calcul de l’iPhone X ? Mieux que les cryptomonnaies pour payer vos courses : vos données personnelles ! Palmer Luckey : l’Elon Musk de l’armée ? Axon, y a le téléphone qui son ! Keecker, Xperia Hello, le futur sera Kawaii. Et le Google Home avec écran ? Enfin : les Instant Apps sur le Play Store. Super Turrican en version Director’s Cut ! Mais sur une machine à 200$... Bonus Guillaume Poggiaspalla : Le coup de gueule des portes ouvertes : SI C’EST OUVERT, N’ENTREZ PAS ! Guillaume Vendé : X Files et la dernière vidéo d’Hygiene Mentale

Welcome to episode 71 of the EdTech Situation Room from October 25, 2017, where technology news meets educational analysis. This week Jason Neiffer (@techsavvyteach) and Wes Fryer (@wfryer) discussed Microsoft's play to offer its universe of applications (including the Edge web browser and the Cortana assistant) on Android phones, upgrade woes with iOS 11, and Coda's efforts to create a new document format merging word processing documents and spreadsheets. Security articles included a shout out to Nicole Perlroth's September 11th interview on cybersecurity on the World Affairs Council podcast, Facebook security issues and the Facebook privacy checkup, and new attacks including "BadRabbit ransomware" and "The Reaper Botnet." Jason Snell's recent article hoping for / predicting a forthcoming Mac Mini update, the incredible learning speed and accomplishments of AlphaGo Zero, SeeSaw's addition of "Activities" to its classroom app, and new/updated clear solar cells were also discussed. Jason finished out the week's articles talking about "containers on Chromebooks." Geeks of the week included discounted Amazon Echos on Woot, the BBEDIT text editor for batch-editing documents, and Storyspheres from Google. Please follow us on Twitter @edtechSR to stay updated, and join us LIVE for a future show at 9 pm Central / 8 pm Mountain. Check all our shownotes on http://edtechsr.com/links

Aquí llega un nuevo episodio lleno de noticias: La seguridad de vuestro WiFi ha caído, tenemos nuevas ondas gravitacionales y muchas noticias sobre sostenibilidad. Además, os contamos cómo el presidente de Volkswagen considera que ellos tienen más responsabilidad social que Tesla. ¡Esperamos que os guste! Noticias El brillo de las ondas gravitacionales creadas al chocar dos estrellas de neutrones, las kilonovas y el origen del oro de la Tierra – http://danielmarin.naukas.com/2017/10/16/detectadas-al-fin-ondas-gravitacionales-provocadas-por-el-choque-de-dos-estrellas-de-neutrones/El protocolo WPA2 ha sido hackeado: la seguridad de las redes WiFi queda comprometida – https://www.xataka.com/seguridad/el-protocolo-wpa2-ha-sido-vulnerado-la-seguridad-de-todas-las-redes-wifi-queda-comprometida Ataque Krack a redes WPA2: así actúa y así puedes protegerte – https://www.xataka.com/seguridad/ataque-krack-a-redes-wpa2-asi-actua-y-asi-puedes-protegerte Las diferencias entre AlphaGo Fan, AlphaGo Lee, AlphaGo Master y AlphaGo Zero – http://francis.naukas.com/2017/10/21/las-diferencias-entre-alphago-fan-alphago-lee-alphago-master-y-alphago-zero/Zero Motorcycles 2018. Hasta 18 kWh y 359 kilómetros de autonomía, sin incremento de precio – https://forococheselectricos.com/2017/10/conoce-las-nuevas-zero-motorcycles-2018-mayor-autonomia-mismo-precio.htmlOpel ordena parar las nuevas reservas del Ampera-e – https://forococheselectricos.com/2017/10/opel-ordena-parar-las-nuevas-reservas-del-ampera-e.htmlEl Ayuntamiento de Murcia dará ayudas a la compra de coches eléctricos, e instalará puntos de recarga públicos – https://forococheselectricos.com/2017/10/el-ayuntamiento-de-murcia-dara-ayudas-la-compra-de-coches-electricos-e-instalara-puntos-de-recarga-publicos.htmlJP Morgan rebaja a la mitad su previsión de producción del Tesla Model 3 para el cuarto trimestre – https://forococheselectricos.com/2017/10/jp-morgan-rebaja-la-mitad-su-prevision-de-produccion-del-tesla-model-3-para-el-cuarto-trimestre.htmlEl presidente de Volkswagen carga duramente contra Tesla. “Sólo fabrican 80.000 unidades, y pierden montones de dinero” – https://forococheselectricos.com/2017/10/el-presidente-de-volkswagen-carga-duramente-contra-tesla-solo-fabrican-80-000-unidades-y-pierden-montones-de-dinero.html Música del episodio Kinematic – PeyoteBlue Giraffe – Emotional theme Podéis encontrarnos en Twitter y en Facebook!

In this episode of the ARCHITECHT AI Show, co-host Signe Brewster speaks with Carnegie Mellon robotics researcher Aaron Steinfeld on the intricacies of human-machine interaction. Steinfeld's work covers a wide range of areas, from autonomous cars to robots that assign blame and credit when working alongside people. In the news segment, co-hosts Brewster, Derrick Harris and Chris Albrecht discuss DeepMind's new AlphaGo Zero system, San Francisco's ban on robot deliveries, and Blade Runner 2049. If your Roomba looked like Ryan Gosling, would you treat if differently?

In this episode of The Next Billion Seconds with Mark Pesce, the latest in A.I. | For thousands of years, amateurs and professionals alike have dedicated themselves to the ancient Chinese board game Go. This week, Google's artificial intelligence group, DeepMind, unveiled a program so smart, it mastered the game in just 3 days, without any human help and only the rules as a guide. In this special episode, Mark explains how AlphaGo Zero didn't stop there, it then hit 2 more milestones, one at 21 days and another no one thought possible in just 40 days. This new benchmark in Artificial intelligence demonstrates how freeing machines from the limits of human knowledge shows us just how smart they can become without us. Find Mark Pesce at: https://twitter.com/mpesce Follow The Show at: https://twitter.com/nextbillionsecs Find PodcastOne Australia on Facebook: https://www.facebook.com/podcastoneau/ Follow PodcastOne Australia on Instagram: https://www.instagram.com/podcastoneau/ Follow PodcastOne Australia on Twitter: https://twitter.com/podcastoneau