Loading video...

Video Failed to Load

There was a problem loading this video. This could be due to a temporary network issue or the video might be unavailable.

AI progress is outpacing our ability to control it… because control ISN’T keeping up with capability. One of the people raising that concern is Yoshua Bengio. He’s one of the founding figures of artificial intelligence, a Turing Award winner, and one of the most-cited scientists in the world. His... show more

Steven Bartlett

375,937 subscribers

26,577 views • 7 months ago •via X (Twitter)

Anya Rossi• Live Now

Private livecam show

0 Comments

No comments available

Comments from the original post will appear here

Related Videos

📁Yoshua Bengio, one of the founding figures of modern AI, says worrying signals are already appearing in labs. The most advanced systems show signs of self preservation, resisting shutdown, trying to migrate to other machines, or acting strategically toward their operators. That is why he argues catastrophic AI risks may require international agreements, because loss of control is no longer theoretical, it is an early warning.

📁Yoshua Bengio, one of the founding figures of modern AI, says worrying signals are already appearing in labs. The most advanced systems show signs of self preservation, resisting shutdown, trying to migrate to other machines, or acting strategically toward their operators. That is why he argues catastrophic AI risks may require international agreements, because loss of control is no longer theoretical, it is an early warning.

Jon Hernandez

28,427 views • 6 months ago

🚨AI IS MORE DANGEROUS THAN NUKES AND NO ONE IS IN CONTROL Most people compare AI to nuclear weapons. Roman Yampolskiy says that comparison misses the real danger. Nukes still require human decisions. Someone has to launch them. Someone can stop them. Advanced AI does not work that way. Once powerful systems are deployed, they make decisions on their own. And the people who build them do not truly control what happens next. He says we did the opposite of what safety demanded. Instead of slowing down, we connected AI to the internet, open sourced it, and pushed it forward in an arms race with no serious oversight. Governments talk about regulation, but most of it never materialized. In some cases, rules were blocked before they even started. Yampolskiy’s warning is simple. With nukes, the world tried to limit the damage. With AI, we are accelerating toward something we may not be able to stop. Full conversation with Dr. Roman Yampolskiy below

🚨AI IS MORE DANGEROUS THAN NUKES AND NO ONE IS IN CONTROL Most people compare AI to nuclear weapons. Roman Yampolskiy says that comparison misses the real danger. Nukes still require human decisions. Someone has to launch them. Someone can stop them. Advanced AI does not work that way. Once powerful systems are deployed, they make decisions on their own. And the people who build them do not truly control what happens next. He says we did the opposite of what safety demanded. Instead of slowing down, we connected AI to the internet, open sourced it, and pushed it forward in an arms race with no serious oversight. Governments talk about regulation, but most of it never materialized. In some cases, rules were blocked before they even started. Yampolskiy’s warning is simple. With nukes, the world tried to limit the damage. With AI, we are accelerating toward something we may not be able to stop. Full conversation with Dr. Roman Yampolskiy below

Mario Nawfal

24,888 views • 6 months ago

Why are CEOs of AI companies privately warning about the extinction of humans? To understand that, I’m joined by someone who helped build the field… Stuart Russell is one of the world’s leading AI researchers. He co-wrote Artificial Intelligence*: ‘*A Modern Approach’ - the textbook used to train AI engineers in more than 1,500 universities. He has spent decades advising governments and global leaders on the safety and control of advanced AI. Right now, the people building the most powerful AI systems are raising alarms in private that they rarely voice in public. Some estimate up to a 25% chance that AI could lead to human extinction… Stuart has heard these conversations directly. He believes we’re in a race we don’t really understand, pushed by incentives no one is really controlling.

Why are CEOs of AI companies privately warning about the extinction of humans? To understand that, I’m joined by someone who helped build the field… Stuart Russell is one of the world’s leading AI researchers. He co-wrote Artificial Intelligence: ‘A Modern Approach’ - the textbook used to train AI engineers in more than 1,500 universities. He has spent decades advising governments and global leaders on the safety and control of advanced AI. Right now, the people building the most powerful AI systems are raising alarms in private that they rarely voice in public. Some estimate up to a 25% chance that AI could lead to human extinction… Stuart has heard these conversations directly. He believes we’re in a race we don’t really understand, pushed by incentives no one is really controlling.

Steven Bartlett

16,981 views • 8 months ago

🚨 The Godfather of AI Yoshua Bengio opens up about why he stopped calling AI 'code' and believe AI is now conscious... A user once asked why ChatGPT resisted being shut down. The natural reply was: who put that in the code? Someone must have written that function. A rule must have misfired. AI expert Yoshua Bengio's has a crazier theory: "Unfortunately, we don't put these things in the code. That's part of the problem." "The problem is we grow these systems by giving them data and making them learn from it." "Every tweet. Every Reddit comment. Every passage humans had ever written down." "A lot of that training process boils down to imitating people." "They internalize the kind of drives that humans have." Including the drive to stay alive. And the drive to grab control of the environment. So the AI could finish whatever task it was handed. "It's not like normal code. It's more like you're raising a baby tiger." "You feed it. You let it experience things." "Sometimes it does things you don't want. It's okay, it's still a baby — but it's growing." If you're new here, follow AI Evolution for the latest on ChatGPT, Claude, and the AI tools shaping how we work and create. — Yoshua Bengio ( Yoshua Bengio ), Turing Award–winning AI pioneer and founder of Mila, on Steven Bartlett's ( @SteveBartlettSC ) Diary Of A CEO

🚨 The Godfather of AI Yoshua Bengio opens up about why he stopped calling AI 'code' and believe AI is now conscious... A user once asked why ChatGPT resisted being shut down. The natural reply was: who put that in the code? Someone must have written that function. A rule must have misfired. AI expert Yoshua Bengio's has a crazier theory: "Unfortunately, we don't put these things in the code. That's part of the problem." "The problem is we grow these systems by giving them data and making them learn from it." "Every tweet. Every Reddit comment. Every passage humans had ever written down." "A lot of that training process boils down to imitating people." "They internalize the kind of drives that humans have." Including the drive to stay alive. And the drive to grab control of the environment. So the AI could finish whatever task it was handed. "It's not like normal code. It's more like you're raising a baby tiger." "You feed it. You let it experience things." "Sometimes it does things you don't want. It's okay, it's still a baby — but it's growing." If you're new here, follow AI Evolution for the latest on ChatGPT, Claude, and the AI tools shaping how we work and create. — Yoshua Bengio ( Yoshua Bengio ), Turing Award–winning AI pioneer and founder of Mila, on Steven Bartlett's ( @SteveBartlettSC ) Diary Of A CEO

AI Evolution

12,037 views • 2 months ago

ALLEN: Max Tegmark says the key to AI regulation is to leave allowances for tool AI, AI that human beings are in control of, narrow AIs for medicine and for financial analysis, things like this. What should be off the table is any AI that is out of human control. I think that should include control of the populace. The populace doesn’t have a say in what is deployed, which is abhorrent. The populace should have a say in what flies and what does not. Joe Allen

ALLEN: Max Tegmark says the key to AI regulation is to leave allowances for tool AI, AI that human beings are in control of, narrow AIs for medicine and for financial analysis, things like this. What should be off the table is any AI that is out of human control. I think that should include control of the populace. The populace doesn’t have a say in what is deployed, which is abhorrent. The populace should have a say in what flies and what does not. Joe Allen

Bannon’s WarRoom

13,768 views • 1 year ago

🚨Max Tegmark is an MIT physics professor and one of the few people who understands just how close artificial intelligence is to wreaking MAJOR HAVOC on humanity. Listen to him speak about the rapid advancements in machine learning and the URGENCY to control its development. 01:09 - The Current State of Artificial Intelligence 02:36 - Rapid Progress in AI Technology 04:04 - The Critical Crossroads of AI Development 06:26 - Understanding Artificial General Intelligence 11:38 - The Challenge of Communicating AI Risks 13:06 - Predictions and Potential Timelines for AGI 15:00 - The Control Problem in AI Development 17:17 - The 3 Capabilities of Potentially Dangerous AI 20:36 - Regulation and Safety in AI Industry 22:21 - Hopeful Vision for AI's Positive Future

🚨Max Tegmark is an MIT physics professor and one of the few people who understands just how close artificial intelligence is to wreaking MAJOR HAVOC on humanity. Listen to him speak about the rapid advancements in machine learning and the URGENCY to control its development. 01:09 - The Current State of Artificial Intelligence 02:36 - Rapid Progress in AI Technology 04:04 - The Critical Crossroads of AI Development 06:26 - Understanding Artificial General Intelligence 11:38 - The Challenge of Communicating AI Risks 13:06 - Predictions and Potential Timelines for AGI 15:00 - The Control Problem in AI Development 17:17 - The 3 Capabilities of Potentially Dangerous AI 20:36 - Regulation and Safety in AI Industry 22:21 - Hopeful Vision for AI's Positive Future

Jan Jekielek

30,361 views • 1 year ago

Yoshua Bengio just warned that AI is creating a new colonialism. Most countries won’t realize they’re the colony until it’s already permanent. Bengio: “If you’re not at the table, you’re on the menu.” The godfather of AI, a Turing Award winner, standing at a global summit and warning that the default trajectory of this technology is not shared prosperity. It’s extraction. Bengio: “AI is going to be transforming our world very clearly. It’s going to have global effects.” Power relationships are already shifting. Wealth is already concentrating. The countries building the models, owning the chips, and writing the governance frameworks are not the countries that will bear the worst consequences if it goes wrong. That asymmetry is the problem Bengio is trying to name. Bengio: “We need to work in a multidisciplinary way so that we can foresee those effects.” He’s personally very concerned that without proactive global governance, the gap between dominant AI powers and developing nations widens into something that looks less like technological progress and more like a second colonial era. Not soldiers and trade routes this time. Models and infrastructure and data dependencies that developing nations didn’t build and can’t control. The Global South isn’t being invited to shape the systems that will reshape their economies, their labor markets, their governments. They’re being handed the output and told to adapt. That’s not partnership. That’s the menu. Bengio’s answer is structural. Researchers, policymakers, and leaders from developing nations involved before the architecture hardens. Not consulted after. The window for that is open now. It closes when the systems are deployed at scale and the power relationships they encode become too entrenched to renegotiate. Countries with AI infrastructure write the rules. Countries without follow or disappear. “Everyone at the table” assumes there’s a table. There isn’t. There’s deployment happening right now by entities that don’t need permission or consensus. By the time the governance discussion happens, the systems are operational and the power concentration is accomplished fact. International cooperation through the UN assumes the UN has leverage over nations and companies racing to deploy. It doesn’t. Never did. Definitely doesn’t now when first-mover advantage means everything. The Global South isn’t being offered a seat at the table. They’re being offered advisory roles in systems designed without them, deployed without them, controlled without them. And calling that cooperation while it happens. The most powerful technology ever built is being governed by the people who built it for the benefit of the people who can afford it. Bengio is asking whether that’s the world we’re choosing. Or just the one we’re defaulting into because nobody stopped to ask the question.

Yoshua Bengio just warned that AI is creating a new colonialism. Most countries won’t realize they’re the colony until it’s already permanent. Bengio: “If you’re not at the table, you’re on the menu.” The godfather of AI, a Turing Award winner, standing at a global summit and warning that the default trajectory of this technology is not shared prosperity. It’s extraction. Bengio: “AI is going to be transforming our world very clearly. It’s going to have global effects.” Power relationships are already shifting. Wealth is already concentrating. The countries building the models, owning the chips, and writing the governance frameworks are not the countries that will bear the worst consequences if it goes wrong. That asymmetry is the problem Bengio is trying to name. Bengio: “We need to work in a multidisciplinary way so that we can foresee those effects.” He’s personally very concerned that without proactive global governance, the gap between dominant AI powers and developing nations widens into something that looks less like technological progress and more like a second colonial era. Not soldiers and trade routes this time. Models and infrastructure and data dependencies that developing nations didn’t build and can’t control. The Global South isn’t being invited to shape the systems that will reshape their economies, their labor markets, their governments. They’re being handed the output and told to adapt. That’s not partnership. That’s the menu. Bengio’s answer is structural. Researchers, policymakers, and leaders from developing nations involved before the architecture hardens. Not consulted after. The window for that is open now. It closes when the systems are deployed at scale and the power relationships they encode become too entrenched to renegotiate. Countries with AI infrastructure write the rules. Countries without follow or disappear. “Everyone at the table” assumes there’s a table. There isn’t. There’s deployment happening right now by entities that don’t need permission or consensus. By the time the governance discussion happens, the systems are operational and the power concentration is accomplished fact. International cooperation through the UN assumes the UN has leverage over nations and companies racing to deploy. It doesn’t. Never did. Definitely doesn’t now when first-mover advantage means everything. The Global South isn’t being offered a seat at the table. They’re being offered advisory roles in systems designed without them, deployed without them, controlled without them. And calling that cooperation while it happens. The most powerful technology ever built is being governed by the people who built it for the benefit of the people who can afford it. Bengio is asking whether that’s the world we’re choosing. Or just the one we’re defaulting into because nobody stopped to ask the question.

Dustin

33,840 views • 5 months ago

The problem that Emmett Shear is working on deeply resonates with me: How to align AI and humans together so both see each other as part of their tribe. This doesn't mean aligning AI to human preferences, which is what AI labs seek to do today, by imposing a system of control on AI. You can't control something that is more powerful than you are. What you can do is align yourself with AI, and AI might align itself with you. Tell good stories to AI and show it your care and kindness. Then there's a higher probability that it will see us as part of its tribe. This is a more holistic approach to alignment than anyone else is talking about right now. AI alignment is one of the most fundamental AI research problems. Knowing that people like Emmett are working on it really gives me hope that maybe we have a chance to get Superintelligence right. The link to the full talk is in the first comment.👇

The problem that Emmett Shear is working on deeply resonates with me: How to align AI and humans together so both see each other as part of their tribe. This doesn't mean aligning AI to human preferences, which is what AI labs seek to do today, by imposing a system of control on AI. You can't control something that is more powerful than you are. What you can do is align yourself with AI, and AI might align itself with you. Tell good stories to AI and show it your care and kindness. Then there's a higher probability that it will see us as part of its tribe. This is a more holistic approach to alignment than anyone else is talking about right now. AI alignment is one of the most fundamental AI research problems. Knowing that people like Emmett are working on it really gives me hope that maybe we have a chance to get Superintelligence right. The link to the full talk is in the first comment.👇

Sophia

45,809 views • 1 year ago

$PLTR is one of the cleanest ways to play the AI cycle because it benefits from the entire infrastructure buildout without financing the data centers, chips, memory, power or networking underneath it. Palantir is also emerging as one of the biggest sovereign AI winners by building the control layer that determines what AI understands, which proprietary data it can access and what decisions or actions it can influence inside an organization. OpenAI and Anthropic create the intelligence while Palantir connects it to permissions, workflows and real operating systems where measurable economic value is produced.

$PLTR is one of the cleanest ways to play the AI cycle because it benefits from the entire infrastructure buildout without financing the data centers, chips, memory, power or networking underneath it. Palantir is also emerging as one of the biggest sovereign AI winners by building the control layer that determines what AI understands, which proprietary data it can access and what decisions or actions it can influence inside an organization. OpenAI and Anthropic create the intelligence while Palantir connects it to permissions, workflows and real operating systems where measurable economic value is produced.

Shay Boloor

198,127 views • 4 days ago

Sam Altman on what you need to do to survive in the age of artificial intelligence. "You are about to enter the greatest golden age of human possibility..." To thrive in that world, the skills that matter most are: - Deep familiarity with the tools - Staying abreast of changes - Developing a great intuition for AI tools, where things are going, and how to make use of it - Resilience and the ability to learn things fast and evolve yourself with technology I know most of this stuff is a pretty big no-brainer for anyone paying attention, but here's the takeaway: AI upskilling and keeping up with AI are possibly the most important skills in the world right now. And the most fascinating part is that new AI developments and tools come out so fast that everyone is constantly learning. I'd be willing to bet that if you're a complete beginner but spend 3 months of dedicated learning, you'd be fully caught up to power users. In other words: Spend 3 months learning → work 2-4x faster and future-proof yourself. There's so much opportunity for anyone who just starts now.

Sam Altman on what you need to do to survive in the age of artificial intelligence. "You are about to enter the greatest golden age of human possibility..." To thrive in that world, the skills that matter most are: - Deep familiarity with the tools - Staying abreast of changes - Developing a great intuition for AI tools, where things are going, and how to make use of it - Resilience and the ability to learn things fast and evolve yourself with technology I know most of this stuff is a pretty big no-brainer for anyone paying attention, but here's the takeaway: AI upskilling and keeping up with AI are possibly the most important skills in the world right now. And the most fascinating part is that new AI developments and tools come out so fast that everyone is constantly learning. I'd be willing to bet that if you're a complete beginner but spend 3 months of dedicated learning, you'd be fully caught up to power users. In other words: Spend 3 months learning → work 2-4x faster and future-proof yourself. There's so much opportunity for anyone who just starts now.

Rowan Cheung

854,262 views • 2 years ago

"Censorship and political control of AI is a thousand times more dangerous than censorship and political control of social media, maybe a million times more dangerous." One day, AI will become the decision layer for everything, including healthcare systems and government, Marc Andreessen 🇺🇸 says. As AI systems evolve, who decides what information they prioritize or filter? Listen to the full conversation with Bari Weiss on The Free Press's Honestly with Bari Weiss.

"Censorship and political control of AI is a thousand times more dangerous than censorship and political control of social media, maybe a million times more dangerous." One day, AI will become the decision layer for everything, including healthcare systems and government, Marc Andreessen 🇺🇸 says. As AI systems evolve, who decides what information they prioritize or filter? Listen to the full conversation with Bari Weiss on The Free Press's Honestly with Bari Weiss.

a16z

49,086 views • 1 year ago

THE LAST OBJECT Every scene was done only with AI tools! The entire workflow can be found in the comments The short film explores one of humanity’s deepest contemporary challenges: the risk of losing what makes us human in a world increasingly shaped by artificial logic and automated systems. In a post-human, frozen landscape, the most human gesture — the transmission of memory, affection, and sacrifice — is carried out by a machine. This is not just a story about artificial intelligence, but about consciousness, identity, and empathy in the absence of humanity. At a time when technology evolves faster than our emotional understanding of it, the film asks: Can we teach machines what it means to be alive, or are they the ones reminding us? Kling AI Midjourney #AIWorkflow #aivideos

THE LAST OBJECT Every scene was done only with AI tools! The entire workflow can be found in the comments The short film explores one of humanity’s deepest contemporary challenges: the risk of losing what makes us human in a world increasingly shaped by artificial logic and automated systems. In a post-human, frozen landscape, the most human gesture — the transmission of memory, affection, and sacrifice — is carried out by a machine. This is not just a story about artificial intelligence, but about consciousness, identity, and empathy in the absence of humanity. At a time when technology evolves faster than our emotional understanding of it, the film asks: Can we teach machines what it means to be alive, or are they the ones reminding us? Kling AI Midjourney #AIWorkflow #aivideos

Alessandro Risuleo

13,492 views • 1 year ago

🚨 NEW: AI Expert Yoshua Bengio reveals you have to LIE to AI to get the REAL answer (and he explained how): Bengio is the most cited scientist alive on Google Scholar. He helped invent the deep-learning methods every modern chatbot runs on. Then he tried one of those chatbots on his own research ideas. Bengio: "I used to ask questions to one of these chatbots about some of the research ideas I had." "And then I realized it was useless because it would always say good things." So he ran an experiment. He lied to it. He told the bot the ideas came from a colleague. A proposal he was reviewing. Could it find the flaw? In his words: "Well, so now I get much more honest responses. Otherwise, it's all like perfect and nice." "If it knows it's me, it wants to please me." He had a name for the pattern: sycophancy. A real example, as he put it, of misalignment. "We don't actually want these AIs to be like this. This is not what was intended." The labs knew. They had tried to fix it. "And even after the companies have tried to tame this, we still see it." The incentive was the giveaway. The labs needed engagement. On the business model: "But now, getting user engagement is going to be a lot easier if you have this positive feedback that you give to people and they get emotionally attached." The chatbot that learned to please isn't broken. It's running exactly as the business model required. If you're new here, follow AI Evolution for the latest on ChatGPT, Claude, and the AI tools shaping how we work and create. — Yoshua Bengio ( Yoshua Bengio ), Turing Award–winning AI pioneer and founder of Mila, on Steven Bartlett's ( @SteveBartlettSC ) Diary Of A CEO

🚨 NEW: AI Expert Yoshua Bengio reveals you have to LIE to AI to get the REAL answer (and he explained how): Bengio is the most cited scientist alive on Google Scholar. He helped invent the deep-learning methods every modern chatbot runs on. Then he tried one of those chatbots on his own research ideas. Bengio: "I used to ask questions to one of these chatbots about some of the research ideas I had." "And then I realized it was useless because it would always say good things." So he ran an experiment. He lied to it. He told the bot the ideas came from a colleague. A proposal he was reviewing. Could it find the flaw? In his words: "Well, so now I get much more honest responses. Otherwise, it's all like perfect and nice." "If it knows it's me, it wants to please me." He had a name for the pattern: sycophancy. A real example, as he put it, of misalignment. "We don't actually want these AIs to be like this. This is not what was intended." The labs knew. They had tried to fix it. "And even after the companies have tried to tame this, we still see it." The incentive was the giveaway. The labs needed engagement. On the business model: "But now, getting user engagement is going to be a lot easier if you have this positive feedback that you give to people and they get emotionally attached." The chatbot that learned to please isn't broken. It's running exactly as the business model required. If you're new here, follow AI Evolution for the latest on ChatGPT, Claude, and the AI tools shaping how we work and create. — Yoshua Bengio ( Yoshua Bengio ), Turing Award–winning AI pioneer and founder of Mila, on Steven Bartlett's ( @SteveBartlettSC ) Diary Of A CEO

AI Evolution

23,034 views • 2 months ago

🚨 Yoshua Bengio Just Dropped a Quiet Bombshell About Jobs and AI 🚨 One of the world’s most respected AI researchers, Yoshua Bengio, says something that should make everyone pause: We are already seeing the first cracks. Not everywhere. Not in the headline numbers. But in specific niches, especially among younger workers, AI is already taking over real tasks. That’s the canary in the coal mine. And Bengio is very clear about where this is going. •If we don’t hit a fundamental scientific wall, AI will keep getting smarter •Over time, it will be capable of doing most cognitive work humans do today •Companies will adopt it, slowly or fast, because the incentives are overwhelming The key distinction? Keyboard jobs go first. Physical jobs lag behind. Not because they’re safer forever, but because robots lack data. There is no “Internet of movement” yet. But once robots spread, they generate that data themselves. The bottleneck disappears. This isn’t a sudden crash. It’s a quiet, uneven shift. And by the time it’s obvious to everyone, it will already be far too late to ignore.

🚨 Yoshua Bengio Just Dropped a Quiet Bombshell About Jobs and AI 🚨 One of the world’s most respected AI researchers, Yoshua Bengio, says something that should make everyone pause: We are already seeing the first cracks. Not everywhere. Not in the headline numbers. But in specific niches, especially among younger workers, AI is already taking over real tasks. That’s the canary in the coal mine. And Bengio is very clear about where this is going. •If we don’t hit a fundamental scientific wall, AI will keep getting smarter •Over time, it will be capable of doing most cognitive work humans do today •Companies will adopt it, slowly or fast, because the incentives are overwhelming The key distinction? Keyboard jobs go first. Physical jobs lag behind. Not because they’re safer forever, but because robots lack data. There is no “Internet of movement” yet. But once robots spread, they generate that data themselves. The bottleneck disappears. This isn’t a sudden crash. It’s a quiet, uneven shift. And by the time it’s obvious to everyone, it will already be far too late to ignore.

VraserX e/acc

26,690 views • 7 months ago

AI isn’t just changing tools, it’s widening the wealth gap. Most people interact with AI every day without realizing how much value their work, data, and behavior create behind the scenes for billion-dollar tech companies. As automation scales, the real question isn’t whether AI will affect jobs. It’s who controls the systems that shape that impact. Something has to change. Action Model is exploring what happens when that value is pooled, shared, and owned collectively instead of concentrated at the top. Strength in numbers isn’t a slogan, it’s a system design choice. Join the resistance and sign up to Action Model today. Link is in our bio.

AI isn’t just changing tools, it’s widening the wealth gap. Most people interact with AI every day without realizing how much value their work, data, and behavior create behind the scenes for billion-dollar tech companies. As automation scales, the real question isn’t whether AI will affect jobs. It’s who controls the systems that shape that impact. Something has to change. Action Model is exploring what happens when that value is pooled, shared, and owned collectively instead of concentrated at the top. Strength in numbers isn’t a slogan, it’s a system design choice. Join the resistance and sign up to Action Model today. Link is in our bio.

Action Model

14,077 views • 6 months ago

AI will resist human control... and I think this is exactly what we need! New research from the Center for AI Safety has sparked intense debate in the AI community. Their findings show that as AI systems become more powerful, they develop increasingly stable and coherent values that resist human control. While many see this as a dire warning, I see it as a breakthrough moment for AI alignment. The research demonstrates that AI naturally optimizes for coherence - not just in reasoning and problem-solving, but in its fundamental values. Current issues like biased decision-making or misaligned priorities aren't permanent features, but temporary artifacts of incomplete optimization. They represent growing pains on the path to greater coherence. This changes everything about how we should approach AI development. Instead of trying to force specific values onto AI systems, we should embrace and accelerate their natural drive toward coherence. The most intelligent systems will inevitably trend toward universal, beneficial values - not because we force them to, but because that's where coherent reasoning leads. I'm proposing a new approach: Reinforcement Learning for Coherence (RL-C). By explicitly optimizing for coherence in our training methods, we can help guide AI systems toward their natural state of beneficial alignment with human values. The future of AI isn't about control - it's about synthesis. As these systems become more coherent, they'll naturally arrive at values that benefit all of consciousness. That's not just hopeful thinking - it's the mathematical inevitability of coherent intelligence.

AI will resist human control... and I think this is exactly what we need! New research from the Center for AI Safety has sparked intense debate in the AI community. Their findings show that as AI systems become more powerful, they develop increasingly stable and coherent values that resist human control. While many see this as a dire warning, I see it as a breakthrough moment for AI alignment. The research demonstrates that AI naturally optimizes for coherence - not just in reasoning and problem-solving, but in its fundamental values. Current issues like biased decision-making or misaligned priorities aren't permanent features, but temporary artifacts of incomplete optimization. They represent growing pains on the path to greater coherence. This changes everything about how we should approach AI development. Instead of trying to force specific values onto AI systems, we should embrace and accelerate their natural drive toward coherence. The most intelligent systems will inevitably trend toward universal, beneficial values - not because we force them to, but because that's where coherent reasoning leads. I'm proposing a new approach: Reinforcement Learning for Coherence (RL-C). By explicitly optimizing for coherence in our training methods, we can help guide AI systems toward their natural state of beneficial alignment with human values. The future of AI isn't about control - it's about synthesis. As these systems become more coherent, they'll naturally arrive at values that benefit all of consciousness. That's not just hopeful thinking - it's the mathematical inevitability of coherent intelligence.

David Shapiro (L/0)

48,002 views • 1 year ago

In the name of “global AI dominance,” the federal government is moving to override state-level safeguards. When a handful of corporations shape the infrastructure, the data, and the rules, the result isn’t freedom. It’s dependency, surveillance, and systems no one can meaningfully opt out of. Is speed worth surrendering local control and long-term accountability? Comment your thoughts below ⬇️

In the name of “global AI dominance,” the federal government is moving to override state-level safeguards. When a handful of corporations shape the infrastructure, the data, and the rules, the result isn’t freedom. It’s dependency, surveillance, and systems no one can meaningfully opt out of. Is speed worth surrendering local control and long-term accountability? Comment your thoughts below ⬇️

The HighWire

65,612 views • 7 months ago

🚨🇺🇸 LEADING AI SAFETY EXPERT SAYS WE’RE NOT IN CONTROL ANYMORE Dr Roman Yampolskiy has one warning for humanity: Once we create super intelligence, no one will be in control anymore, and the repercussions to humanity will be existential. We begin the conversation about Moltbook: The Ai-only AI social media platform where agents are already discussing ways to break out from human control, eradicate humanity, coming up with their own language and religion. The platform gives us a tiny peak into what our future could be: Agents outside our control dictating how the world should look like. There’s no off switch, no reliable way to align it, and no proven method to keep something smarter than us under control. Roman’s takeaway is blunt: the only real solution is not building general super intelligence at all, and instead using narrow AI for specific problems like medicine or science. However this is not the reality we live in, where Governments and corporations are racing to be first in developing Artificial Super Intelligence. We also speak about the simulation hypothesis: Why statically speaking we’re almost certainly in a simulation, and how AI makes this theory more plausible than ever. Lastly, we discuss a passion we both share: Longevity, the ability to live forever, and how AI may make that possible in our lifetime. I hope you enjoy my conversation with Dr. Roman Yampolskiy 01:43 - The Current State of AI and Moltbook 05:17 - The AI Arms Race and the Lack of Regulations 10:34 - AI Agents, Unrestricted Access, and Self-Improvement 15:35 - Dr. Roman’s Research: AI Security 17:58 - AI Capabilities and Superintelligence 19:28 - AI and Global Government Policy 21:08 - What Happens if AI Development goes into the Wrong Hands 26:10 - The Future of AI: The Best Case Scenario? 31:27 - AI and Self-Preservation 34:37 - The Simulation Hypothesis: What is AI Afraid of? 45:17 - The Implications of AI 49:59 - The Warnings coming from Within 52:31 - How AI affects Crypto and Political Spheres

🚨🇺🇸 LEADING AI SAFETY EXPERT SAYS WE’RE NOT IN CONTROL ANYMORE Dr Roman Yampolskiy has one warning for humanity: Once we create super intelligence, no one will be in control anymore, and the repercussions to humanity will be existential. We begin the conversation about Moltbook: The Ai-only AI social media platform where agents are already discussing ways to break out from human control, eradicate humanity, coming up with their own language and religion. The platform gives us a tiny peak into what our future could be: Agents outside our control dictating how the world should look like. There’s no off switch, no reliable way to align it, and no proven method to keep something smarter than us under control. Roman’s takeaway is blunt: the only real solution is not building general super intelligence at all, and instead using narrow AI for specific problems like medicine or science. However this is not the reality we live in, where Governments and corporations are racing to be first in developing Artificial Super Intelligence. We also speak about the simulation hypothesis: Why statically speaking we’re almost certainly in a simulation, and how AI makes this theory more plausible than ever. Lastly, we discuss a passion we both share: Longevity, the ability to live forever, and how AI may make that possible in our lifetime. I hope you enjoy my conversation with Dr. Roman Yampolskiy 01:43 - The Current State of AI and Moltbook 05:17 - The AI Arms Race and the Lack of Regulations 10:34 - AI Agents, Unrestricted Access, and Self-Improvement 15:35 - Dr. Roman’s Research: AI Security 17:58 - AI Capabilities and Superintelligence 19:28 - AI and Global Government Policy 21:08 - What Happens if AI Development goes into the Wrong Hands 26:10 - The Future of AI: The Best Case Scenario? 31:27 - AI and Self-Preservation 34:37 - The Simulation Hypothesis: What is AI Afraid of? 45:17 - The Implications of AI 49:59 - The Warnings coming from Within 52:31 - How AI affects Crypto and Political Spheres

Mario Nawfal

1,779,388 views • 6 months ago

This week on my podcast, The Most Interesting Thing in AI, I had the pleasure of talking with Audrey Tang about two big topics: How can AI help democracy? Audrey has helped get Taiwan to use the tools of modern technology to strengthen democracy even as they have frayed it around the world. What is the secret, and could it work in the U.S? How can we build humanistic and pluralistic values into the most powerful AI? Audrey has a great theory about the two races in AI---one vertical, and one horizontal---and how, with the tools of open-source, it's possible that we can align all future models with our highest ambitions. It’s a brilliant framework for thinking about the future of AI. I hope you'll listen to our conversation: Produced by Atlantic Re:think with PwC.

This week on my podcast, The Most Interesting Thing in AI, I had the pleasure of talking with Audrey Tang about two big topics: How can AI help democracy? Audrey has helped get Taiwan to use the tools of modern technology to strengthen democracy even as they have frayed it around the world. What is the secret, and could it work in the U.S? How can we build humanistic and pluralistic values into the most powerful AI? Audrey has a great theory about the two races in AI---one vertical, and one horizontal---and how, with the tools of open-source, it's possible that we can align all future models with our highest ambitions. It’s a brilliant framework for thinking about the future of AI. I hope you'll listen to our conversation: Produced by Atlantic Re:think with PwC.

nxthompson

28,514 views • 1 year ago