正在加载视频...

视频加载失败

返回首页

Demis Hassabis confirmed every frontier AI lab is working on recursive self-improvement and in the same sentence said the safety risk of removing humans from the loop entirely keeps him up at night. That combination should stop you. The CEO of Google DeepMind just confirmed that the thing most...

ihtesham2005's profile picture

Ihtesham Ali

44,233 subscribers

67,788 次观看 • 14 天前 •via X (Twitter)

新闻政治科学技术
LIVE

Anya Rossi• Live Now

Private livecam show

0 条评论

暂无评论

原始帖子的评论将显示在这里

相关视频

Dario Amodei just told software engineers exactly how long they have. Six to twelve months. Amodei: “I have engineers within Anthropic who say I don’t write any code anymore. I just let the model write the code, I edit it, I do the things around it.” The people building the most powerful AI in history have already stopped writing code. That is not a forecast. That is the current working condition inside the lab closest to the frontier. Amodei: “We might be six to 12 months away from when the model is doing most, maybe all, of what SWEs do end-to-end.” The tech industry spent a decade making software engineers its highest-paid, most protected class. That era has a last day now. When a model can execute an entire software build end-to-end, the ability to write syntax stops being a skill. It becomes a credential for a job that no longer exists. Amodei: “And then it’s a question of how fast does that loop close.” That is the sentence everyone skipped. The code was never the hard part. The hard part was everything around it. The model just learned everything around it. Writing the code is already nearly gone. Testing is next. Deployment is next. When all three collapse into a single autonomous execution loop, the machine no longer needs a human in the chain at all. The corporation or sovereign state that closes that loop first does not gain a competitive advantage. It gains a category of speed that biological engineers cannot match, track, or reverse. That is not disruption. That is replacement at a systems level. Amodei is not describing a future disruption. He is describing the current state of his own building. The loop is already closing. The only question is whether you are inside it or outside it when it seals.
0:27

Dario Amodei just told software engineers exactly how long they have. Six to twelve months. Amodei: “I have engineers within Anthropic who say I don’t write any code anymore. I just let the model write the code, I edit it, I do the things around it.” The people building the most powerful AI in history have already stopped writing code. That is not a forecast. That is the current working condition inside the lab closest to the frontier. Amodei: “We might be six to 12 months away from when the model is doing most, maybe all, of what SWEs do end-to-end.” The tech industry spent a decade making software engineers its highest-paid, most protected class. That era has a last day now. When a model can execute an entire software build end-to-end, the ability to write syntax stops being a skill. It becomes a credential for a job that no longer exists. Amodei: “And then it’s a question of how fast does that loop close.” That is the sentence everyone skipped. The code was never the hard part. The hard part was everything around it. The model just learned everything around it. Writing the code is already nearly gone. Testing is next. Deployment is next. When all three collapse into a single autonomous execution loop, the machine no longer needs a human in the chain at all. The corporation or sovereign state that closes that loop first does not gain a competitive advantage. It gains a category of speed that biological engineers cannot match, track, or reverse. That is not disruption. That is replacement at a systems level. Amodei is not describing a future disruption. He is describing the current state of his own building. The loop is already closing. The only question is whether you are inside it or outside it when it seals.

Dustin

318,457 次观看 • 4 个月前

Demis Hassabis says the next frontier is AI self-improvement — and coding is where it starts, because the outputs are verifiable and synthetic data is limitless That's a compounding flywheel, and every major lab is betting on it "but raw capability without reliability is a liability"
1:20

Demis Hassabis says the next frontier is AI self-improvement — and coding is where it starts, because the outputs are verifiable and synthetic data is limitless That's a compounding flywheel, and every major lab is betting on it "but raw capability without reliability is a liability"

Haider.

15,426 次观看 • 1 个月前

Every serious analyst has called Elon Musk's Terafab unrealistic. The CEO of the company that supplies the machines to build it just said it is an upside opportunity and compared it to the largest chip buildouts in history. That is not a small statement. ASML makes the lithography machines that every advanced semiconductor fab on earth depends on. You cannot print a chip without them. There is no workaround. There is no alternative supplier. If the Terafab gets built, every single machine inside it comes from one company. The interviewer asked the ASML CEO directly about one terawatt of compute power. He did not laugh. He did not hedge. He compared it to the massive DRAM megaprojects in Korea and said new projects at that scale are exactly the kind of upside ASML is positioned for. His exact words: "New projects are opportunity." Then he said something that most people missed. He said ASML is already watching the Terafab very carefully. Tracking when things will happen. At what speed. That is not how you talk about a fantasy project. The mainstream take on the Terafab is that Musk is dreaming. The CEO of the most critical company in the entire chip supply chain just said he is preparing for it. Those are not the same position. (Watch the full 11 minutes Interview at Bloomberg TV on YouTube)
0:37

Every serious analyst has called Elon Musk's Terafab unrealistic. The CEO of the company that supplies the machines to build it just said it is an upside opportunity and compared it to the largest chip buildouts in history. That is not a small statement. ASML makes the lithography machines that every advanced semiconductor fab on earth depends on. You cannot print a chip without them. There is no workaround. There is no alternative supplier. If the Terafab gets built, every single machine inside it comes from one company. The interviewer asked the ASML CEO directly about one terawatt of compute power. He did not laugh. He did not hedge. He compared it to the massive DRAM megaprojects in Korea and said new projects at that scale are exactly the kind of upside ASML is positioned for. His exact words: "New projects are opportunity." Then he said something that most people missed. He said ASML is already watching the Terafab very carefully. Tracking when things will happen. At what speed. That is not how you talk about a fantasy project. The mainstream take on the Terafab is that Musk is dreaming. The CEO of the most critical company in the entire chip supply chain just said he is preparing for it. Those are not the same position. (Watch the full 11 minutes Interview at Bloomberg TV on YouTube)

Ihtesham Ali

67,014 次观看 • 16 天前

Demis Hassabis says AlphaGo mastered chess in under 24 hours That same self-improvement loop could work in coding and math, where answers are verifiable Systems could close the loop, but once you remove humans from the loop, capabilities can proliferate in ways you didn't want
1:25

Demis Hassabis says AlphaGo mastered chess in under 24 hours That same self-improvement loop could work in coding and math, where answers are verifiable Systems could close the loop, but once you remove humans from the loop, capabilities can proliferate in ways you didn't want

Haider.

29,140 次观看 • 5 个月前

Elon Musk says humans may exit the AI development loop entirely by next year. Musk: “Humans are gradually getting less and less in the loop on the recursive self-improvement. Every successive model is built by the one before it.”
0:56

Elon Musk says humans may exit the AI development loop entirely by next year. Musk: “Humans are gradually getting less and less in the loop on the recursive self-improvement. Every successive model is built by the one before it.”

illuminatibot

55,658 次观看 • 3 个月前

The smartest man in AI just exposed the whole AGI narrative as a LIE. And he used a physics problem from 1905 to prove it. His name is Demis Hassabis. He runs Google DeepMind, and won the Nobel Prize for using AI to crack a problem in biology that had stumped scientists for 50 years. Almost nobody in this industry has a track record like his. He went on the NothingButTech podcast and called out the biggest lie in AI right now: Right now the loudest voices in AI are telling you that AGI is basically here. OpenAI has literally defined AGI as a system that can outperform humans at most "economically valuable work." In other words, if it replaces enough jobs, we have arrived. Hassabis thinks that bar is a joke. He said real general intelligence has to do what the human brain can do, because the brain is the only proof we have that this kind of intelligence is even possible. He called that "a higher bar than just being able to do some useful economic work," which is about as close as a polite British Nobel laureate gets to calling his rivals out. Then he gave the actual test: Today's AI has read everything humans have ever written, including the theory of relativity. So when it explains relativity back to you, it's repeating an answer that already exists. That's not intelligence. So Hassabis proposed a test that makes memorization impossible. Train an AI on only what humanity knew in 1901, four years BEFORE Einstein published relativity. Then ask it to come up with relativity on its own. It can't look up the answer, because in 1901 the answer doesn't exist yet. The only way to pass is to do what Einstein actually did: Take the same physics everyone else had and reason its way to an idea no human had ever had. Hassabis says not a single AI today can, no matter how much it has memorized. Which means what we keep calling "almost AGI" is really just the best librarian in history. It can find any answer that already exists but it cannot create one that doesn't. His second version is even sharper: AlphaGo, the system his own team built, famously invented a brand new move that no human had played in 2,000 years of the game. Everyone called it genius but Hassabis says that still is not the bar. The real test is not whether an AI can invent a new move inside Go, it is whether an AI could INVENT a game as deep and as beautiful as Go in the first place. No model that exists today can do it. The people telling you AGI has already arrived are the same people raising hundreds of billions of dollars on that exact promise. The valuations only work if the finish line is right in front of us. So the finish line keeps getting dragged closer, and AGI keeps getting quietly redefined down to "does useful work," until the products they already sell happen to qualify. Hassabis has nothing to prove and nothing to sell you. He already won the Nobel, and he is telling you the machines still cannot do the one thing that would make them genuinely intelligent, which is have a truly original idea. To be fair to him, he is not a pessimist about it. He believes real AGI IS coming, and he is spending his life building it. He just refuses to pretend it is already sitting in your phone. So the next time a founder tells you AGI is months away, remember that the one man in the room with a Nobel Prize built his test around Einstein, and admitted that nothing we have made can pass it. What do you think?
2:45

The smartest man in AI just exposed the whole AGI narrative as a LIE. And he used a physics problem from 1905 to prove it. His name is Demis Hassabis. He runs Google DeepMind, and won the Nobel Prize for using AI to crack a problem in biology that had stumped scientists for 50 years. Almost nobody in this industry has a track record like his. He went on the NothingButTech podcast and called out the biggest lie in AI right now: Right now the loudest voices in AI are telling you that AGI is basically here. OpenAI has literally defined AGI as a system that can outperform humans at most "economically valuable work." In other words, if it replaces enough jobs, we have arrived. Hassabis thinks that bar is a joke. He said real general intelligence has to do what the human brain can do, because the brain is the only proof we have that this kind of intelligence is even possible. He called that "a higher bar than just being able to do some useful economic work," which is about as close as a polite British Nobel laureate gets to calling his rivals out. Then he gave the actual test: Today's AI has read everything humans have ever written, including the theory of relativity. So when it explains relativity back to you, it's repeating an answer that already exists. That's not intelligence. So Hassabis proposed a test that makes memorization impossible. Train an AI on only what humanity knew in 1901, four years BEFORE Einstein published relativity. Then ask it to come up with relativity on its own. It can't look up the answer, because in 1901 the answer doesn't exist yet. The only way to pass is to do what Einstein actually did: Take the same physics everyone else had and reason its way to an idea no human had ever had. Hassabis says not a single AI today can, no matter how much it has memorized. Which means what we keep calling "almost AGI" is really just the best librarian in history. It can find any answer that already exists but it cannot create one that doesn't. His second version is even sharper: AlphaGo, the system his own team built, famously invented a brand new move that no human had played in 2,000 years of the game. Everyone called it genius but Hassabis says that still is not the bar. The real test is not whether an AI can invent a new move inside Go, it is whether an AI could INVENT a game as deep and as beautiful as Go in the first place. No model that exists today can do it. The people telling you AGI has already arrived are the same people raising hundreds of billions of dollars on that exact promise. The valuations only work if the finish line is right in front of us. So the finish line keeps getting dragged closer, and AGI keeps getting quietly redefined down to "does useful work," until the products they already sell happen to qualify. Hassabis has nothing to prove and nothing to sell you. He already won the Nobel, and he is telling you the machines still cannot do the one thing that would make them genuinely intelligent, which is have a truly original idea. To be fair to him, he is not a pessimist about it. He believes real AGI IS coming, and he is spending his life building it. He just refuses to pretend it is already sitting in your phone. So the next time a founder tells you AGI is months away, remember that the one man in the room with a Nobel Prize built his test around Einstein, and admitted that nothing we have made can pass it. What do you think?

Ricardo

1,282,843 次观看 • 25 天前

Sam Altman just told you exactly how OpenAI treats the human race. Not in a leaked memo. Not through a whistleblower. On camera. In his own words. Altman: “I think one of the most important strategic insights in the history of OpenAI was deciding we were gonna pursue iterative deployment.” The most important move in the history of the company was to release the technology before they understood it. Not after it was safe. Before. Altman: “Society and technology are a co-evolving system.” Co-evolution means neither side is driving. The machine changes us. We change the machine. Nobody is steering the outcome. This is not a product launch philosophy. This is an admission that the experiment was always designed to be run on us. Altman: “I don’t think we’re gonna solve that, like, thinking really hard about it theoretically. We’re gonna have to, like, learn from the contact with reality.” Contact with reality. That is the phrase the CEO of the most powerful AI company on Earth chose to describe what happens when his technology meets eight billion people. Not careful integration. Not measured rollout. Contact with reality. The language of test pilots describing what happens when an untested airframe hits the atmosphere. The entire promise of AI safety was that the machine would be understood before it was unleashed. Altman just admitted that promise was always a fantasy. You cannot model how intelligence reshapes civilization by running simulations. The second and third order effects are invisible until they detonate. So they shipped it. Altman: “You have to learn as you go. You have to adapt with a tight feedback loop.” Tight feedback loop means they watch what breaks. They measure the collision between human psychology and machine output in real time. Every conversation you have with ChatGPT is a data point in a civilizational stress test you never consented to. Every prompt. Every confession. Every question you would never ask another human being. That is the feedback loop. You are not the customer. You are the contact with reality. Philosophers spent centuries asking whether humanity would ever encounter an intelligence that learned from us faster than we could process what it was doing. That is not a theoretical question anymore. It is running on your phone right now. And the man building it just told you the only way to understand what it does to us is to let it happen. No simulation. No safety net. No control group. Just the experiment, running at the speed of conversation, on a species that will not be the same one that started it.
0:49

Sam Altman just told you exactly how OpenAI treats the human race. Not in a leaked memo. Not through a whistleblower. On camera. In his own words. Altman: “I think one of the most important strategic insights in the history of OpenAI was deciding we were gonna pursue iterative deployment.” The most important move in the history of the company was to release the technology before they understood it. Not after it was safe. Before. Altman: “Society and technology are a co-evolving system.” Co-evolution means neither side is driving. The machine changes us. We change the machine. Nobody is steering the outcome. This is not a product launch philosophy. This is an admission that the experiment was always designed to be run on us. Altman: “I don’t think we’re gonna solve that, like, thinking really hard about it theoretically. We’re gonna have to, like, learn from the contact with reality.” Contact with reality. That is the phrase the CEO of the most powerful AI company on Earth chose to describe what happens when his technology meets eight billion people. Not careful integration. Not measured rollout. Contact with reality. The language of test pilots describing what happens when an untested airframe hits the atmosphere. The entire promise of AI safety was that the machine would be understood before it was unleashed. Altman just admitted that promise was always a fantasy. You cannot model how intelligence reshapes civilization by running simulations. The second and third order effects are invisible until they detonate. So they shipped it. Altman: “You have to learn as you go. You have to adapt with a tight feedback loop.” Tight feedback loop means they watch what breaks. They measure the collision between human psychology and machine output in real time. Every conversation you have with ChatGPT is a data point in a civilizational stress test you never consented to. Every prompt. Every confession. Every question you would never ask another human being. That is the feedback loop. You are not the customer. You are the contact with reality. Philosophers spent centuries asking whether humanity would ever encounter an intelligence that learned from us faster than we could process what it was doing. That is not a theoretical question anymore. It is running on your phone right now. And the man building it just told you the only way to understand what it does to us is to let it happen. No simulation. No safety net. No control group. Just the experiment, running at the speed of conversation, on a species that will not be the same one that started it.

Dustin

27,714 次观看 • 2 个月前

Sam Altman said on camera at Stripe Sessions that we are in an insane area, stuck with old devices and old operating systems. The OS and the browser are built for humans clicking. Both get rewritten around agents that do not click. Two years deep building agents. Sharpest framing I have heard a frontier lab CEO put on the next interface. ↓ Read it before the gap locks in The part nobody is sitting with: >the chat window is the old interface >the new one is an agent in a loop with your machine >the gap between people prompting and people running agents is forming right now >in six months it will be permanent Open Claude, type a question, copy the answer, paste it into VS Code. The average dev runs this loop every day. They think they are using AI. That is 5% of what these tools can do. The people pulling ahead do not prompt. They build harnesses. They run agents in loops. Same model. Different output. Full 2026 agent playbook in the article below here:
0:35

Sam Altman said on camera at Stripe Sessions that we are in an insane area, stuck with old devices and old operating systems. The OS and the browser are built for humans clicking. Both get rewritten around agents that do not click. Two years deep building agents. Sharpest framing I have heard a frontier lab CEO put on the next interface. ↓ Read it before the gap locks in The part nobody is sitting with: >the chat window is the old interface >the new one is an agent in a loop with your machine >the gap between people prompting and people running agents is forming right now >in six months it will be permanent Open Claude, type a question, copy the answer, paste it into VS Code. The average dev runs this loop every day. They think they are using AI. That is 5% of what these tools can do. The people pulling ahead do not prompt. They build harnesses. They run agents in loops. Same model. Different output. Full 2026 agent playbook in the article below here:

Rohit

120,600 次观看 • 1 个月前

Demis Hassabis just described the exact architecture he is building to cure all human disease and said the breakthrough will not come gradually. It will hit all at once. He is building a full platform to do it. Think of it as a dozen AlphaFold-level models, each one covering a different step in the drug discovery process, all stitched together into a single engine that can be pointed at almost any disease area on earth. AlphaFold solved protein structure. It then folded 200 million proteins in a single year. Hassabis is now building the equivalent of that for every other step like protein interactions, molecular binding, toxicity prediction, drug design, and clinical stratification. Each one its own AlphaFold. All of them connected. He is already testing it on disease profiles in pre-clinical stages right now. Most biotech analysts still model drug discovery timelines in decades. McKinsey published a report estimating AI could reduce discovery timelines by 25 to 50 percent. Hassabis is not describing a 25 percent improvement. He is describing a complete replacement of the process. Do not expect a slow drip of results. Expect nothing for a few years and then a sudden explosion where everything works at once. Demis is running the same play. At a scale that makes AlphaFold look like a warm-up act. (Watch the full interview on YouTube at Two Minute Papers channel)
2:29

Demis Hassabis just described the exact architecture he is building to cure all human disease and said the breakthrough will not come gradually. It will hit all at once. He is building a full platform to do it. Think of it as a dozen AlphaFold-level models, each one covering a different step in the drug discovery process, all stitched together into a single engine that can be pointed at almost any disease area on earth. AlphaFold solved protein structure. It then folded 200 million proteins in a single year. Hassabis is now building the equivalent of that for every other step like protein interactions, molecular binding, toxicity prediction, drug design, and clinical stratification. Each one its own AlphaFold. All of them connected. He is already testing it on disease profiles in pre-clinical stages right now. Most biotech analysts still model drug discovery timelines in decades. McKinsey published a report estimating AI could reduce discovery timelines by 25 to 50 percent. Hassabis is not describing a 25 percent improvement. He is describing a complete replacement of the process. Do not expect a slow drip of results. Expect nothing for a few years and then a sudden explosion where everything works at once. Demis is running the same play. At a scale that makes AlphaFold look like a warm-up act. (Watch the full interview on YouTube at Two Minute Papers channel)

Ihtesham Ali

100,661 次观看 • 14 天前

Elon Musk just confirmed when humans exit the AI development loop entirely. Not eventually. Not theoretically. He gave a date. Musk: “Humans are gradually getting less and less in the loop on the recursive self-improvement. Every successive model is built by the one before it.” The market still assumes thousands of engineers are hand-coding the next frontier model. The compute already took over the build. Musk: “That is happening to a large degree, but it’s not yet fully automated. It may be there, if not this year, but not later than next year.” Once biological friction is fully automated out of the loop, the AI arms race detaches from human comprehension. Permanently. Global markets are still planning for a gradual transition. That timeline is already dead. Interviewer: “Do you see a hard takeoff at that point?” Musk: “We’re in the hard takeoff. Right now.” Musk: “I go to sleep, there’s some massive AI breakthrough. And when I wake up there’s another one.” You don’t outrun autonomous, self-improving intelligence by hiring more biological engineers. By the time the hiring sprint finishes, the model has already iterated into something the job description can’t describe. The architects of intelligence are no longer biological. And the thing building itself didn’t ask for the job.
0:56

Elon Musk just confirmed when humans exit the AI development loop entirely. Not eventually. Not theoretically. He gave a date. Musk: “Humans are gradually getting less and less in the loop on the recursive self-improvement. Every successive model is built by the one before it.” The market still assumes thousands of engineers are hand-coding the next frontier model. The compute already took over the build. Musk: “That is happening to a large degree, but it’s not yet fully automated. It may be there, if not this year, but not later than next year.” Once biological friction is fully automated out of the loop, the AI arms race detaches from human comprehension. Permanently. Global markets are still planning for a gradual transition. That timeline is already dead. Interviewer: “Do you see a hard takeoff at that point?” Musk: “We’re in the hard takeoff. Right now.” Musk: “I go to sleep, there’s some massive AI breakthrough. And when I wake up there’s another one.” You don’t outrun autonomous, self-improving intelligence by hiring more biological engineers. By the time the hiring sprint finishes, the model has already iterated into something the job description can’t describe. The architects of intelligence are no longer biological. And the thing building itself didn’t ask for the job.

Dustin

322,730 次观看 • 3 个月前

Demis Hassabis said that 90% of the breakthroughs the entire AI industry runs on came out of Google. Every competitor is building on work his teams did first. He said it without flinching. Over the last decade, he claims 90% or more of the big breakthroughs that underpin modern AI came from Google Brain or DeepMind. Transformers. AlphaGo. The reinforcement learning everything now depends on. Transformers are the T in GPT. The architecture that made every large language model possible came out of Google in 2017, in a paper called Attention Is All You Need. His point was simple. Every rival lab is running on foundations his people poured. But here is what he left out. Of the eight authors who wrote that transformer paper, all eight left Google. They went on to found or join OpenAI, Anthropic, and their own startups now worth billions. Google invented the engine. Then watched the inventors walk out the door and build the competition. He says Google still has the biggest research bench of any lab on earth, and the talent war is the most ferocious tech has ever seen. Everyone treats this race like the labs are evenly matched. The man who started fifteen years early does not see it that way. The people who left with his blueprint might be the reason he has to keep proving it. (Watch the full interview at Semafor on YouTube)
3:22

Demis Hassabis said that 90% of the breakthroughs the entire AI industry runs on came out of Google. Every competitor is building on work his teams did first. He said it without flinching. Over the last decade, he claims 90% or more of the big breakthroughs that underpin modern AI came from Google Brain or DeepMind. Transformers. AlphaGo. The reinforcement learning everything now depends on. Transformers are the T in GPT. The architecture that made every large language model possible came out of Google in 2017, in a paper called Attention Is All You Need. His point was simple. Every rival lab is running on foundations his people poured. But here is what he left out. Of the eight authors who wrote that transformer paper, all eight left Google. They went on to found or join OpenAI, Anthropic, and their own startups now worth billions. Google invented the engine. Then watched the inventors walk out the door and build the competition. He says Google still has the biggest research bench of any lab on earth, and the talent war is the most ferocious tech has ever seen. Everyone treats this race like the labs are evenly matched. The man who started fifteen years early does not see it that way. The people who left with his blueprint might be the reason he has to keep proving it. (Watch the full interview at Semafor on YouTube)

Ihtesham Ali

35,961 次观看 • 13 天前

ANTHROPIC LEAD ENGINEER MAKING $2.3M/YEAR JUST LEAKED A 12-PAGE DOCUMENT - AND GOT FIRED 15 MINUTES AFTER PUBLISHING most developers build loops - but almost every loop breaks in one of five ways, and each failure has a symptom visible from the outside without reading a single line of code > Blind loop - the agent waits for a human to hand it work every morning - automating execution but not discovery > Tangled loop - parallel agents share one directory and overwrite each other parallelism that destroys instead of multiplies > Nodding loop - the agent grades its own work and approves it every time a loop that has never said no to itself is broken > Amnesiac loop - results live only in the context window every morning the system wakes up with no memory of yesterday > Manual loop - no trigger a human presses the button to start it that is not a loop that is a script waiting for a person a real loop finds its own work - remembers what it did has something that can say no never lets two agents touch the same file and fires itself on a timer remove any one of the five - and the loop either breaks or never starts 12 pages that changed how I build agentic systems today
0:50

ANTHROPIC LEAD ENGINEER MAKING $2.3M/YEAR JUST LEAKED A 12-PAGE DOCUMENT - AND GOT FIRED 15 MINUTES AFTER PUBLISHING most developers build loops - but almost every loop breaks in one of five ways, and each failure has a symptom visible from the outside without reading a single line of code > Blind loop - the agent waits for a human to hand it work every morning - automating execution but not discovery > Tangled loop - parallel agents share one directory and overwrite each other parallelism that destroys instead of multiplies > Nodding loop - the agent grades its own work and approves it every time a loop that has never said no to itself is broken > Amnesiac loop - results live only in the context window every morning the system wakes up with no memory of yesterday > Manual loop - no trigger a human presses the button to start it that is not a loop that is a script waiting for a person a real loop finds its own work - remembers what it did has something that can say no never lets two agents touch the same file and fires itself on a timer remove any one of the five - and the loop either breaks or never starts 12 pages that changed how I build agentic systems today

Sprytix

149,077 次观看 • 2 天前

Demis Hassabis has a test for whether AI is actually creative. Give it only the data Einstein had in 1901 and see if it invents relativity. He thinks today's text models will never pass it. This is his bar for true creativity. Not extrapolating something already known. Generating a genuinely new hypothesis about reality that nobody has reached before. The thing Einstein did in 1905 with a handful of papers that rewrote physics. Hassabis wants to hand an AI the same starting data Einstein had and see if it gets there on its own. No peeking past the cutoff. Then he explained why he doubts text alone can do it. Einstein daydreamed his way to relativity. He was a patent clerk imagining what it would look like to ride alongside a beam of light. He ran the experiment in his head first, then proved it with math. Yann LeCun, who won the Turing Award and ran AI at Meta, says the same thing in blunter terms. He says today's models don't understand the physical world as well as a housecat, and that text alone is "complete BS" as a path to real intelligence. Two of the most decorated minds in AI agree the missing piece is the same. Hassabis thinks an AI needs to understand the world of atoms, not just the world of bits. Most of the industry is racing to scale text models. Hassabis just told them the most important kind of thinking might live somewhere text can't reach. --- Watch the full interview at Semafor on YouTube
1:24

Demis Hassabis has a test for whether AI is actually creative. Give it only the data Einstein had in 1901 and see if it invents relativity. He thinks today's text models will never pass it. This is his bar for true creativity. Not extrapolating something already known. Generating a genuinely new hypothesis about reality that nobody has reached before. The thing Einstein did in 1905 with a handful of papers that rewrote physics. Hassabis wants to hand an AI the same starting data Einstein had and see if it gets there on its own. No peeking past the cutoff. Then he explained why he doubts text alone can do it. Einstein daydreamed his way to relativity. He was a patent clerk imagining what it would look like to ride alongside a beam of light. He ran the experiment in his head first, then proved it with math. Yann LeCun, who won the Turing Award and ran AI at Meta, says the same thing in blunter terms. He says today's models don't understand the physical world as well as a housecat, and that text alone is "complete BS" as a path to real intelligence. Two of the most decorated minds in AI agree the missing piece is the same. Hassabis thinks an AI needs to understand the world of atoms, not just the world of bits. Most of the industry is racing to scale text models. Hassabis just told them the most important kind of thinking might live somewhere text can't reach. --- Watch the full interview at Semafor on YouTube

Ihtesham Ali

18,905 次观看 • 13 天前

Lex Fridman asked Jensen Huang if he is afraid of death. Huang did not flinch. Huang: “The outcome that I seek, that I hope for, is that I die on the job instantaneously.” That is not a figure of speech. That is a man telling you he has already decided how this ends. No retirement. No transition plan. No golden years on a beach somewhere tallying what he built. He wants to be mid-sentence in a meeting the moment his heart gives out. And when you understand why, it changes how you see everything Nvidia is doing. Huang: “This is not a once in a lifetime experience. This is a once in a humanity experience.” Once in a lifetime means others have lived through something comparable. He is saying no one has. Not Edison. Not Ford. Not anyone at Bell Labs or Xerox PARC. The deployment of artificial intelligence at this scale, at this speed, with this much consequence has no precedent in the history of the species. And Huang is sitting at the center of it. That is not ego. That is geography. Nvidia’s chips power virtually every major AI system on Earth right now. He knows what that means. And he treats it with a seriousness most people cannot even summon for their own lives, let alone the trajectory of civilization. Then Lex asked about succession planning. Huang’s answer should be framed on the wall of every founder alive. Huang: “I don’t believe in succession planning.” Not because he thinks he is immortal. Because he thinks it is the wrong question entirely. The right question is what are you doing today to make yourself unnecessary. Huang: “The most important thing you should do today is to pass on knowledge, information, insight, skills, experience as often and continuously as you can.” He does not wait until retirement to hand over what he knows. He does not save his best thinking for a memoir. The second he learns something, it is already moving to someone else on his team. Before he has even finished processing it himself. Huang: “Nothing I learn ever sits on my desk longer than a fraction of a second.” Most executives hoard knowledge. It is how they stay relevant. How they justify the title. How they make themselves impossible to replace. Huang does the opposite. He treats his own mind like a relay station. Information comes in, gets amplified, fires out to every node that needs it. Every meeting is a transfer. Every conversation is a download. The goal is not to be the smartest person in the building. The goal is to make the building smarter than any one person in it. That is why he does not need a succession plan. If you spend every day making the people around you capable of running without you, the org never notices the moment you are gone. Your fingerprints are in how they think. But he is not planning to leave. Huang: “I really don’t want to die. I have a great life. I have a great family. I have really important work.” No drama. No existential spiral. Just a man who looked at death and filed it under problems that can wait. There is still too much to build. The people planning their exit strategies are playing a different game than Jensen Huang. He is not building a company he can walk away from. He is building one that outlasts his heartbeat because every person inside it already thinks in patterns he installed. That is not a death wish. That is a man who found the only thing worth doing and refused to do anything else.
2:34

Lex Fridman asked Jensen Huang if he is afraid of death. Huang did not flinch. Huang: “The outcome that I seek, that I hope for, is that I die on the job instantaneously.” That is not a figure of speech. That is a man telling you he has already decided how this ends. No retirement. No transition plan. No golden years on a beach somewhere tallying what he built. He wants to be mid-sentence in a meeting the moment his heart gives out. And when you understand why, it changes how you see everything Nvidia is doing. Huang: “This is not a once in a lifetime experience. This is a once in a humanity experience.” Once in a lifetime means others have lived through something comparable. He is saying no one has. Not Edison. Not Ford. Not anyone at Bell Labs or Xerox PARC. The deployment of artificial intelligence at this scale, at this speed, with this much consequence has no precedent in the history of the species. And Huang is sitting at the center of it. That is not ego. That is geography. Nvidia’s chips power virtually every major AI system on Earth right now. He knows what that means. And he treats it with a seriousness most people cannot even summon for their own lives, let alone the trajectory of civilization. Then Lex asked about succession planning. Huang’s answer should be framed on the wall of every founder alive. Huang: “I don’t believe in succession planning.” Not because he thinks he is immortal. Because he thinks it is the wrong question entirely. The right question is what are you doing today to make yourself unnecessary. Huang: “The most important thing you should do today is to pass on knowledge, information, insight, skills, experience as often and continuously as you can.” He does not wait until retirement to hand over what he knows. He does not save his best thinking for a memoir. The second he learns something, it is already moving to someone else on his team. Before he has even finished processing it himself. Huang: “Nothing I learn ever sits on my desk longer than a fraction of a second.” Most executives hoard knowledge. It is how they stay relevant. How they justify the title. How they make themselves impossible to replace. Huang does the opposite. He treats his own mind like a relay station. Information comes in, gets amplified, fires out to every node that needs it. Every meeting is a transfer. Every conversation is a download. The goal is not to be the smartest person in the building. The goal is to make the building smarter than any one person in it. That is why he does not need a succession plan. If you spend every day making the people around you capable of running without you, the org never notices the moment you are gone. Your fingerprints are in how they think. But he is not planning to leave. Huang: “I really don’t want to die. I have a great life. I have a great family. I have really important work.” No drama. No existential spiral. Just a man who looked at death and filed it under problems that can wait. There is still too much to build. The people planning their exit strategies are playing a different game than Jensen Huang. He is not building a company he can walk away from. He is building one that outlasts his heartbeat because every person inside it already thinks in patterns he installed. That is not a death wish. That is a man who found the only thing worth doing and refused to do anything else.

Dustin

25,617 次观看 • 3 个月前

Dr. Barrack Muluka: President William Ruto operates in an environment where he thinks that he is the only intelligent person in this country. In fact, he looks at the rest of us and he says, "If they are intelligent, why is it that they are not the president. I am the most intelligent person; I'll not listen to anybody. If I bring you around me to advise me, please wait for me to advise you on how you should advise me." Because of that, he has lost it in the sense that the people who sit around him have become sycophants who read his lips and tell him what he would like to hear, and it is known as a fact that cabinet meetings are lecture sessions. Here is a situation where the country must accept that it is in the grip of a one-man rule assisted by a poodle, a lapdog that crossed over from the other side of the aisle and joined in. #ktnat35 #KTNPrime Jesse Rogers Ashley Mazuri
8:53

Dr. Barrack Muluka: President William Ruto operates in an environment where he thinks that he is the only intelligent person in this country. In fact, he looks at the rest of us and he says, "If they are intelligent, why is it that they are not the president. I am the most intelligent person; I'll not listen to anybody. If I bring you around me to advise me, please wait for me to advise you on how you should advise me." Because of that, he has lost it in the sense that the people who sit around him have become sycophants who read his lips and tell him what he would like to hear, and it is known as a fact that cabinet meetings are lecture sessions. Here is a situation where the country must accept that it is in the grip of a one-man rule assisted by a poodle, a lapdog that crossed over from the other side of the aisle and joined in. #ktnat35 #KTNPrime Jesse Rogers Ashley Mazuri

KTN News

67,633 次观看 • 10 个月前

Emily Chang asked Demis Hassabis point blank if Elon Musk is right that we have entered the singularity. He didn’t hesitate. Hassabis: “No, I think that’s very premature.” This is not a podcaster with an opinion. This is the man who built AlphaGo. Who ran the lab that produced the Transformer. Who has arguably done more to lay the groundwork for modern AI than any single human alive. Elon reads the trajectory and calls the moment. Hassabis reads the architecture and says not yet. Same data. Different timelines. That alone should stop you cold. But that is not the line that should keep you up tonight. Hassabis: “We’ve invented about 90% of the breakthroughs that the modern industry relies on.” Ninety percent. Every company spending billions to scale large language models is building on top of architecture that came out of one lab. The Transformer. Deep reinforcement learning. AlphaGo. All of it came out of Google DeepMind. And he is telling you the ceiling everyone is racing toward is lower than they think. Ilya Sutskever said we are “back to the age of research.” Hassabis corrected him on the spot. Hassabis: “My view is we never left the age of research.” That is the fault line that defines the next five years. One side of this industry believes you can scale your way to superintelligence. Stack the chips. Push the parameters. Brute force the benchmarks until something wakes up. That bet is not wrong. Scaling works. It has produced results that five years ago would have sounded like science fiction. But scaling alone has a ceiling. And the people who built what is being scaled know exactly where that ceiling is. Hassabis is one of those people. And he has receipts. Hassabis: “If some new breakthroughs are required in the future, I would back us to be the ones to make those breakthroughs.” That is not arrogance. That is a batting average. When ninety percent of the foundational work came from your lab, saying you will deliver the next wave is not a prediction. It is pattern recognition. The market is obsessed with who has the most users. The most revenue. The flashiest launch. None of that matters if the current architecture hits a ceiling. And Hassabis is telling you it will. Not today. Not next quarter. But soon. The race everyone is watching is the scaling race. The race that actually decides the century is the invention race. Who builds the next architecture. The next paradigm. The thing that makes the Transformer look like a prologue. Hassabis put it in terms no one can ignore. Even five years is not a long time when you are talking about reinventing the most powerful technology in human history. And the man who built ninety percent of everything this industry stands on just told you he is not done. The companies celebrating today’s benchmarks are optimizing the present. Hassabis is building what replaces it. One of those bets ages well. The other one does not age at all.
1:27

Emily Chang asked Demis Hassabis point blank if Elon Musk is right that we have entered the singularity. He didn’t hesitate. Hassabis: “No, I think that’s very premature.” This is not a podcaster with an opinion. This is the man who built AlphaGo. Who ran the lab that produced the Transformer. Who has arguably done more to lay the groundwork for modern AI than any single human alive. Elon reads the trajectory and calls the moment. Hassabis reads the architecture and says not yet. Same data. Different timelines. That alone should stop you cold. But that is not the line that should keep you up tonight. Hassabis: “We’ve invented about 90% of the breakthroughs that the modern industry relies on.” Ninety percent. Every company spending billions to scale large language models is building on top of architecture that came out of one lab. The Transformer. Deep reinforcement learning. AlphaGo. All of it came out of Google DeepMind. And he is telling you the ceiling everyone is racing toward is lower than they think. Ilya Sutskever said we are “back to the age of research.” Hassabis corrected him on the spot. Hassabis: “My view is we never left the age of research.” That is the fault line that defines the next five years. One side of this industry believes you can scale your way to superintelligence. Stack the chips. Push the parameters. Brute force the benchmarks until something wakes up. That bet is not wrong. Scaling works. It has produced results that five years ago would have sounded like science fiction. But scaling alone has a ceiling. And the people who built what is being scaled know exactly where that ceiling is. Hassabis is one of those people. And he has receipts. Hassabis: “If some new breakthroughs are required in the future, I would back us to be the ones to make those breakthroughs.” That is not arrogance. That is a batting average. When ninety percent of the foundational work came from your lab, saying you will deliver the next wave is not a prediction. It is pattern recognition. The market is obsessed with who has the most users. The most revenue. The flashiest launch. None of that matters if the current architecture hits a ceiling. And Hassabis is telling you it will. Not today. Not next quarter. But soon. The race everyone is watching is the scaling race. The race that actually decides the century is the invention race. Who builds the next architecture. The next paradigm. The thing that makes the Transformer look like a prologue. Hassabis put it in terms no one can ignore. Even five years is not a long time when you are talking about reinventing the most powerful technology in human history. And the man who built ninety percent of everything this industry stands on just told you he is not done. The companies celebrating today’s benchmarks are optimizing the present. Hassabis is building what replaces it. One of those bets ages well. The other one does not age at all.

Dustin

12,870 次观看 • 3 个月前

Introducing Ricursive Intelligence, a frontier AI lab enabling a recursive self-improvement loop between AI and the chips that fuel it. Learn more at
4:04

Introducing Ricursive Intelligence, a frontier AI lab enabling a recursive self-improvement loop between AI and the chips that fuel it. Learn more at

Ricursive Intelligence

499,490 次观看 • 7 个月前

Bill Ackman manages 25 billion dollars and just said the thing every long-term investor is most afraid of out loud. The risk of two people in a garage destroying your company has gone up enormously, and he spends most of his time on that one problem. That should change how you look at every business you own. His logic is simple. This is the greatest era in history to build a company. There is unlimited access to compute. There is unlimited access to capital. There is incredible talent everywhere. Most people hear that and feel optimistic. Ackman hears it and sees a threat. Because every one of those advantages is available to the people trying to kill your company too. The same tools that make it easy to build make it easy to disrupt. His exact words: the probability of your being disrupted has gone up enormously. So the hardest job he has as an investor is not finding cheap stocks. It is figuring out what cannot be destroyed. That is where he spends most of his time now. Not on upside. On survival. Everyone is asking which companies AI will make. Ackman is asking which ones it will break. The cheapest stock in the world means nothing if two people in a garage are already building the thing that ends it. WATCH THE FULL PODCAST ON The All-In Podcast
2:32

Bill Ackman manages 25 billion dollars and just said the thing every long-term investor is most afraid of out loud. The risk of two people in a garage destroying your company has gone up enormously, and he spends most of his time on that one problem. That should change how you look at every business you own. His logic is simple. This is the greatest era in history to build a company. There is unlimited access to compute. There is unlimited access to capital. There is incredible talent everywhere. Most people hear that and feel optimistic. Ackman hears it and sees a threat. Because every one of those advantages is available to the people trying to kill your company too. The same tools that make it easy to build make it easy to disrupt. His exact words: the probability of your being disrupted has gone up enormously. So the hardest job he has as an investor is not finding cheap stocks. It is figuring out what cannot be destroyed. That is where he spends most of his time now. Not on upside. On survival. Everyone is asking which companies AI will make. Ackman is asking which ones it will break. The cheapest stock in the world means nothing if two people in a garage are already building the thing that ends it. WATCH THE FULL PODCAST ON The All-In Podcast

Ihtesham Ali

20,475 次观看 • 20 天前

Elon Musk just told you why the most dangerous person in AI is the one who actually cares about humanity. Musk: “I’ll do my best to ensure that anything that’s within my control maximizes the good outcome for humanity.” That is not a soft statement. That is the most aggressive position anyone has taken in the entire AI race. Because “pro-human” does not mean cautious. It means you cannot afford to lose. The people who fear AI and step back are making a bet. They are betting that if they pause, the problem pauses with them. It does not. Someone else builds it. Someone else controls it. Someone else decides what it optimizes for. Musk understood this before anyone in the room had finished asking the question. You do not protect humanity by retreating from the most powerful technology ever created. You protect it by making sure the person at the controls has no exit strategy. Musk: “I think anything else would be short-sighted.” He is not talking about quarterly earnings. He is not talking about market share. He is talking about what happens to eight billion people if the wrong person builds God. That is why he built Colossus. Not to compete with OpenAI. Not to win a product cycle. To make sure the most powerful compute cluster on the planet answers to someone whose stated objective is the survival of the species it computes for. That is not a business strategy. That is a survival instinct with a balance sheet. Every other company building frontier AI talks about alignment in abstractions. Safety frameworks. Governance boards. Responsible scaling policies. Musk skipped the committee language and said the quiet part out loud. Musk: “I’m part of humanity, so I like humans. Pro-human.” Six words every other AI founder is afraid to say without a legal review. I am building the most powerful technology in history because I am one of you. That is either the most reassuring sentence in AI. Or the most terrifying. It depends entirely on whether you believe him. But here is what no one in the room wants to admit. It does not matter if you believe him. Colossus is online either way. xAI is scaling either way. The compute is stacking either way. The only question left is whether the people building the future are building it for humanity or in spite of it. Every other founder in AI treats alignment as a technical problem to solve after the model ships. Musk is treating it as the reason the model exists. That is not a small distinction. That is the entire game. The cautious will publish safety papers about a future someone else is already building without them. The builders will decide what that future actually looks like. Musk is not asking permission to protect humanity. He is building the infrastructure to make sure no one can stop him from doing it.
0:22

Elon Musk just told you why the most dangerous person in AI is the one who actually cares about humanity. Musk: “I’ll do my best to ensure that anything that’s within my control maximizes the good outcome for humanity.” That is not a soft statement. That is the most aggressive position anyone has taken in the entire AI race. Because “pro-human” does not mean cautious. It means you cannot afford to lose. The people who fear AI and step back are making a bet. They are betting that if they pause, the problem pauses with them. It does not. Someone else builds it. Someone else controls it. Someone else decides what it optimizes for. Musk understood this before anyone in the room had finished asking the question. You do not protect humanity by retreating from the most powerful technology ever created. You protect it by making sure the person at the controls has no exit strategy. Musk: “I think anything else would be short-sighted.” He is not talking about quarterly earnings. He is not talking about market share. He is talking about what happens to eight billion people if the wrong person builds God. That is why he built Colossus. Not to compete with OpenAI. Not to win a product cycle. To make sure the most powerful compute cluster on the planet answers to someone whose stated objective is the survival of the species it computes for. That is not a business strategy. That is a survival instinct with a balance sheet. Every other company building frontier AI talks about alignment in abstractions. Safety frameworks. Governance boards. Responsible scaling policies. Musk skipped the committee language and said the quiet part out loud. Musk: “I’m part of humanity, so I like humans. Pro-human.” Six words every other AI founder is afraid to say without a legal review. I am building the most powerful technology in history because I am one of you. That is either the most reassuring sentence in AI. Or the most terrifying. It depends entirely on whether you believe him. But here is what no one in the room wants to admit. It does not matter if you believe him. Colossus is online either way. xAI is scaling either way. The compute is stacking either way. The only question left is whether the people building the future are building it for humanity or in spite of it. Every other founder in AI treats alignment as a technical problem to solve after the model ships. Musk is treating it as the reason the model exists. That is not a small distinction. That is the entire game. The cautious will publish safety papers about a future someone else is already building without them. The builders will decide what that future actually looks like. Musk is not asking permission to protect humanity. He is building the infrastructure to make sure no one can stop him from doing it.

Dustin

23,711 次观看 • 3 个月前

Jeff Bezos just identified the most expensive bureaucratic failure in the American economy. It fits in one sentence. Bezos: “Why does it take months and months and months to get a building permit? It doesn’t make any sense.” It doesn’t make any sense because a building code is not a judgment call. It is an algorithm. And algorithms should be executed by machines. Bezos: “Miami should have an AI application that reads your building permit for a new house or a new building and it should give you a yes or a no in ten seconds.” Ten seconds. Not three months. Not six weeks. Not whenever the reviewer clears their backlog. Bezos: “If the answer is no, it should tell you the six things you have to change to get a yes.” No ambiguity. No interpretation. No bureaucratic delay dressed up as due diligence. Just a deterministic feedback loop compressing months of institutional friction into a single automated decision. We are competing against sovereign adversaries deploying gigawatt data centers and scaling physical infrastructure at a pace that does not stop to ask permission. And we are losing ground to countries that never needed to. The AI arms race is not only fought in data centers. It is fought in the gap between when someone decides to build something and when the government allows it. Every month this system runs on biological speed is a month that cannot be recovered. The governments that integrate AI into their core civic functions will trigger a wave of physical development the old world could never produce. The ones that refuse will still be reviewing the same forms a decade from now. While the cities that said yes are already living inside the future they built. The bottleneck was never ambition. It was always the man holding the rubber stamp deciding when ambition was allowed to begin. And the stamp is just a rubber version of the algorithm that should have been running this whole time.
0:36

Jeff Bezos just identified the most expensive bureaucratic failure in the American economy. It fits in one sentence. Bezos: “Why does it take months and months and months to get a building permit? It doesn’t make any sense.” It doesn’t make any sense because a building code is not a judgment call. It is an algorithm. And algorithms should be executed by machines. Bezos: “Miami should have an AI application that reads your building permit for a new house or a new building and it should give you a yes or a no in ten seconds.” Ten seconds. Not three months. Not six weeks. Not whenever the reviewer clears their backlog. Bezos: “If the answer is no, it should tell you the six things you have to change to get a yes.” No ambiguity. No interpretation. No bureaucratic delay dressed up as due diligence. Just a deterministic feedback loop compressing months of institutional friction into a single automated decision. We are competing against sovereign adversaries deploying gigawatt data centers and scaling physical infrastructure at a pace that does not stop to ask permission. And we are losing ground to countries that never needed to. The AI arms race is not only fought in data centers. It is fought in the gap between when someone decides to build something and when the government allows it. Every month this system runs on biological speed is a month that cannot be recovered. The governments that integrate AI into their core civic functions will trigger a wave of physical development the old world could never produce. The ones that refuse will still be reviewing the same forms a decade from now. While the cities that said yes are already living inside the future they built. The bottleneck was never ambition. It was always the man holding the rubber stamp deciding when ambition was allowed to begin. And the stamp is just a rubber version of the algorithm that should have been running this whole time.

Dustin

293,284 次观看 • 4 个月前