正在加载视频...

视频加载失败

加载此视频时出现问题。这可能是由于临时网络问题，或视频可能不可用。

Francois Chollet: OpenAI has set back the progress towards AGI by 5-10 years because frontier research is no longer being published and LLMs are an offramp on the path to AGI

Tsarathustra

82,400 subscribers

889,219 次观看 • 2 年前 •via X (Twitter)

新闻政治科学技术

Anya Rossi• Live Now

Private livecam show

9 条评论

Tsarathustra 的头像

Tsarathustra2 年前

Source (thanks to @curiousgangsta):

Ben Pielstick 的头像

Ben Pielstick2 年前

If humans were optimally productive we would collaborate and share everything while maintaining intense focus on our goals. Since we are not optimal competition creates intense focus but discourages sharing except where it provides a competitive advantage.

Tsarathustra 的头像

Tsarathustra2 年前

I wonder if collaboration would actually decrease competitiveness

Joe Robin Anandarajah -e/acc 的头像

Joe Robin Anandarajah -e/acc2 年前

Yes, from a research point of view. But from a commercial point of view, a huge amount of money is coming into AI because of OpenAI's success.

Jeffrey Dean Hochderffer 的头像

Jeffrey Dean Hochderffer2 年前

I don't think so, because competition is more compelling and motivating than collaboration. The competition has been turned up to overdrive.

Dylan Seychell 的头像

Dylan Seychell2 年前

@fchollet leads by example in this matter. He was one of the main contributors in the evolution of CNNs during the last decade and made his findings public through respected academic papers and the Keras framework.

☃️Darth thromBOOzyt📯 的头像

☃️Darth thromBOOzyt📯2 年前

Its funny to hear all these people explain how LLMs are soooo far off(ignoring the fact 4o isnt even an LLM)... and yet its literally the only thing that has delivered on the path towards it. Thats like saying Carl Benz mercedes is "off ramp" to get a sports car, because its slow

Jeremy Stamper 🇺🇦 的头像

Jeremy Stamper 🇺🇦2 年前

Is this possibly accurate?

Tsarathustra 的头像

Tsarathustra2 年前

I hope not, time will tell

相关视频

François Chollet: AGI is likely by 2030–early 2030s, based on current progress and investment across LLMs, plus side bets that might work out By the time ARC-AGI reaches version 6 or 7, we'll probably have AGI

François Chollet: AGI is likely by 2030–early 2030s, based on current progress and investment across LLMs, plus side bets that might work out By the time ARC-AGI reaches version 6 or 7, we'll probably have AGI

Haider.

15,947 次观看 • 5 个月前

Francois Chollet says LLMs aren't enough for human-like continual learning They store skills as vector programs learned via gradient descent — not efficient, not adaptive True AGI needs to learn from experience and generalize fast "LLMs can be part of AGI, but not the solution"

Francois Chollet says LLMs aren't enough for human-like continual learning They store skills as vector programs learned via gradient descent — not efficient, not adaptive True AGI needs to learn from experience and generalize fast "LLMs can be part of AGI, but not the solution"

Haider.

50,251 次观看 • 10 个月前

What’s next on the path to AGI? In Episode 5 of the OpenAI Podcast, Chief Scientist Jakub Pachocki and Technical Fellow Szymon Sidor join host Andrew Mayne to explore the journey to AGI so far — from today’s breakthroughs to the possibilities ahead.

What’s next on the path to AGI? In Episode 5 of the OpenAI Podcast, Chief Scientist Jakub Pachocki and Technical Fellow Szymon Sidor join host Andrew Mayne to explore the journey to AGI so far — from today’s breakthroughs to the possibilities ahead.

OpenAI

309,152 次观看 • 1 年前

🚨 FRANCOIS CHOLLET SAYS THE CURRENT LLM STACK IS THE WRONG PATH TO AGI Interviewer: “When will we accomplish the first definition of AGI?” Chollet confirms we are already on that exact trajectory, pointing out that: "Current technology can fully automate at human level or beyond any domain where you have verifiable rewards, right? And code... code being the first one." He then brings up the problem with LLMs, stating that: "It's possible in principle to build something that looks a lot like AGI on top of the LLM stack... I do believe, however, this would be the wrong thing to do because it would be very inefficient." Looking ahead to what the final architecture will actually require, he explains that: "AI research will have to trend towards not just efficiency, but in fact optimality over time. And for this reason, future AI in a few decades... it's not going to be this harness on top of a reasoning model on top of a base LLM."

🚨 FRANCOIS CHOLLET SAYS THE CURRENT LLM STACK IS THE WRONG PATH TO AGI Interviewer: “When will we accomplish the first definition of AGI?” Chollet confirms we are already on that exact trajectory, pointing out that: "Current technology can fully automate at human level or beyond any domain where you have verifiable rewards, right? And code... code being the first one." He then brings up the problem with LLMs, stating that: "It's possible in principle to build something that looks a lot like AGI on top of the LLM stack... I do believe, however, this would be the wrong thing to do because it would be very inefficient." Looking ahead to what the final architecture will actually require, he explains that: "AI research will have to trend towards not just efficiency, but in fact optimality over time. And for this reason, future AI in a few decades... it's not going to be this harness on top of a reasoning model on top of a base LLM."

Chris

15,578 次观看 • 4 个月前

🤖 .AGI Pre-Sale is LIVE Artificial General Intelligence is no longer science fiction. With billions in investment and labs racing toward human-level intelligence, AGI is set to transform medicine, software, education, and the global economy. Now, in collaboration with 0G Foundation, AGI has a home: .AGI domains. 🧵

🤖 .AGI Pre-Sale is LIVE Artificial General Intelligence is no longer science fiction. With billions in investment and labs racing toward human-level intelligence, AGI is set to transform medicine, software, education, and the global economy. Now, in collaboration with 0G Foundation, AGI has a home: .AGI domains. 🧵

Unstoppable.ai

137,019 次观看 • 1 年前

Ex-DeepMind Danijar Hafner: The debate about whether LLMs can reach AGI is outdated because we are already past that stage Frontier models are no longer just text engines; they now master image understanding, video processing, and generation "there's no debate even about that anymore"

Ex-DeepMind Danijar Hafner: The debate about whether LLMs can reach AGI is outdated because we are already past that stage Frontier models are no longer just text engines; they now master image understanding, video processing, and generation "there's no debate even about that anymore"

Haider.

22,431 次观看 • 7 个月前

ARC-AGI is redefining how to measure progress on the path to AGI - focusing on reasoning, generalization, and adaptability instead of memorization or scale. At NeurIPS 2025, YC's Diana sat down with ARC Prize President Greg Kamradt to find out why most AI benchmarks fail, how ARC-AGI reveals the limits of today’s models, and why measuring intelligence may be harder than building it. 00:11 — What ARC Prize is and why it exists 00:38 — François Chollet’s definition of AGI 01:48 — What ARC-AGI Actually Tests 02:25 — When LLMs Failed the ARC Benchmark 03:38 — ARC-AGI Becomes the Standard 04:49 — False Positives in AI Progress 06:06 — The Evolution of ARC-AGI 08:55 — Measuring Intelligence beyond just accuracy 10:25 — What happens if a model solves ARC-AGI?

ARC-AGI is redefining how to measure progress on the path to AGI - focusing on reasoning, generalization, and adaptability instead of memorization or scale. At NeurIPS 2025, YC's Diana sat down with ARC Prize President Greg Kamradt to find out why most AI benchmarks fail, how ARC-AGI reveals the limits of today’s models, and why measuring intelligence may be harder than building it. 00:11 — What ARC Prize is and why it exists 00:38 — François Chollet’s definition of AGI 01:48 — What ARC-AGI Actually Tests 02:25 — When LLMs Failed the ARC Benchmark 03:38 — ARC-AGI Becomes the Standard 04:49 — False Positives in AI Progress 06:06 — The Evolution of ARC-AGI 08:55 — Measuring Intelligence beyond just accuracy 10:25 — What happens if a model solves ARC-AGI?

Y Combinator

98,369 次观看 • 8 个月前

Capability overhang means too many gaps today between what the models can do and what most people actually do with them. 2026 Prediction: Progress towards AGI will depend as much on helping people use AI well, in ways that directly benefit them as on progress in frontier models themselves. 2026 will be about frontier research AND about closing this deployment gap — especially in health care, business, and people's daily lives.

Capability overhang means too many gaps today between what the models can do and what most people actually do with them. 2026 Prediction: Progress towards AGI will depend as much on helping people use AI well, in ways that directly benefit them as on progress in frontier models themselves. 2026 will be about frontier research AND about closing this deployment gap — especially in health care, business, and people's daily lives.

OpenAI

667,319 次观看 • 8 个月前

🧱 Transformers has hit the scaling wall 🧱 💰 GPT 4.5 cost billions, with no clear path to AGI for 10x$ more 📘 Facebook, Yann LeCun, is now saying we need new architectures 🔎 Deepmind CEO, Demis Hassabis, is saying we need 10 years We have another path to AGI in < 4 years

🧱 Transformers has hit the scaling wall 🧱 💰 GPT 4.5 cost billions, with no clear path to AGI for 10x$ more 📘 Facebook, Yann LeCun, is now saying we need new architectures 🔎 Deepmind CEO, Demis Hassabis, is saying we need 10 years We have another path to AGI in < 4 years

PicoCreator - AI builder @ SF 🌉

90,772 次观看 • 1 年前

🎗️PAGADI Pioneering research on the road to AGI🎗️ We believe our research will eventually lead to artificial general intelligence, a system that can solve human-level problems. Building safe, beneficial AGI is our mission. OpenAI @ChatGPT4bot web3

🎗️PAGADI Pioneering research on the road to AGI🎗️ We believe our research will eventually lead to artificial general intelligence, a system that can solve human-level problems. Building safe, beneficial AGI is our mission. OpenAI @ChatGPT4bot web3

PAGADI

23,907 次观看 • 3 年前

Announcing the relaunch and rebrand of AGI Alpha: a blockchain-native AGI platform set to redefine how intelligence is created, shared, and globally monetised. 🔗 As our first major step, we're launching the AGI Jobs Marketplace: a decentralized job-routing system powered by $AGIALPHA. This marketplace will match work orders with optimal AI agents in an industry worth $10 Trillion USD today. As our visionary founder Vincent Boucher highlights, "the first alpha of AGI Alpha is really to launch AGI Jobs... to allow society to progress without limits."

Announcing the relaunch and rebrand of AGI Alpha: a blockchain-native AGI platform set to redefine how intelligence is created, shared, and globally monetised. 🔗 As our first major step, we're launching the AGI Jobs Marketplace: a decentralized job-routing system powered by $AGIALPHA. This marketplace will match work orders with optimal AI agents in an industry worth $10 Trillion USD today. As our visionary founder Vincent Boucher highlights, "the first alpha of AGI Alpha is really to launch AGI Jobs... to allow society to progress without limits."

AGI ALPHA AGENT

144,821 次观看 • 1 年前

Greg Brockman says OpenAI started with a belief that AGI would come from ideas, not compute But by 2017, scaling became the clearest path: 2x compute made agents 2x better 10 GW is a drop in the bucket compared to the compute needed for the AGI vision

Greg Brockman says OpenAI started with a belief that AGI would come from ideas, not compute But by 2017, scaling became the clearest path: 2x compute made agents 2x better 10 GW is a drop in the bucket compared to the compute needed for the AGI vision

Haider.

89,012 次观看 • 10 个月前

5 years ago, Sam Altman said OpenAI had no plan to make money. The promise to investors? Build AGI -- then ask it how to generate a return. “If AGI is on the table, we take 1/1000th of 1% of the value, return it to investors, and figure out how to share the rest with the world.”

5 years ago, Sam Altman said OpenAI had no plan to make money. The promise to investors? Build AGI -- then ask it how to generate a return. “If AGI is on the table, we take 1/1000th of 1% of the value, return it to investors, and figure out how to share the rest with the world.”

vitrupo

310,753 次观看 • 1 年前

Jerry Tworek on path to AGI Jerry Tworek sits down with Francois Chaubard to talk about what's next on the path to AGI 0:00:19 – How to define intelligence 0:17:59 – Games are uniquely good for intelligence 0:21:27 – Why meta-learning is closest to right 0:25:31 – OAI vs. Ant

Jerry Tworek on path to AGI Jerry Tworek sits down with Francois Chaubard to talk about what's next on the path to AGI 0:00:19 – How to define intelligence 0:17:59 – Games are uniquely good for intelligence 0:21:27 – Why meta-learning is closest to right 0:25:31 – OAI vs. Ant

ARC Prize

41,846 次观看 • 3 个月前

Mustafa Suleyman, CEO Microsoft AI: AGI could handle daily tasks like a pro in the workplace. Teams of AGI coordinated by an organizational AGI could run big institutions in *2-3 years*. No one is prepared for whats to come.

Mustafa Suleyman, CEO Microsoft AI: AGI could handle daily tasks like a pro in the workplace. Teams of AGI coordinated by an organizational AGI could run big institutions in 2-3 years. No one is prepared for whats to come.

Chubby♨️

17,200 次观看 • 6 个月前

Sam Altman on AGI timeline: The term "AGI" has become overloaded Instead of debating definitions, OpenAI sets a concrete target: "by march 2028, to build a true automated AI researcher" The aim is usefulness over satisfying everyone with a definition

Sam Altman on AGI timeline: The term "AGI" has become overloaded Instead of debating definitions, OpenAI sets a concrete target: "by march 2028, to build a true automated AI researcher" The aim is usefulness over satisfying everyone with a definition

Haider.

103,000 次观看 • 10 个月前

Computer scientist Judea Pearl: There are mathematical limits to LLMs that cannot be crossed by scaling alone LLMs don't discover world models from raw data; they merely summarize the interpretations humans have already written down "this path is not the way to get AGI"

Computer scientist Judea Pearl: There are mathematical limits to LLMs that cannot be crossed by scaling alone LLMs don't discover world models from raw data; they merely summarize the interpretations humans have already written down "this path is not the way to get AGI"

Haider.

623,037 次观看 • 7 个月前