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。

📁 Yann LeCun explains that LLMs work well when problems are symbolic, like math, code or chess, where searching through known sequences is enough. But the real world does not work that way. Physical action, planning and understanding what is possible require continuous intuition, which LLMs lack. Manipulating symbols... show more

Jon Hernandez

23,739 subscribers

42,623 次观看 • 5 个月前 •via X (Twitter)

教育科学技术

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

Yann LeCun says language isn’t intelligence. Predicting text doesn’t mean understanding reality. The real world is messy, physical, and causal and today’s LLMs barely touch that. The next leap is Physical AI: world models, cause and effect, real planning. Do you think LLMs can evolve into this, or do we need a completely new architecture?

Yann LeCun says language isn’t intelligence. Predicting text doesn’t mean understanding reality. The real world is messy, physical, and causal and today’s LLMs barely touch that. The next leap is Physical AI: world models, cause and effect, real planning. Do you think LLMs can evolve into this, or do we need a completely new architecture?

VraserX e/acc

76,154 次观看 • 4 个月前

📁 Geoffrey Hinton, a pioneer of deep learning, says LLMs do understand what they say. He explains that understanding language is not translating it into an internal symbolic code. In neural networks, words act like Lego blocks that combine to form meaning. It is not translation, it is structure. That is what real language understanding looks like.

📁 Geoffrey Hinton, a pioneer of deep learning, says LLMs do understand what they say. He explains that understanding language is not translating it into an internal symbolic code. In neural networks, words act like Lego blocks that combine to form meaning. It is not translation, it is structure. That is what real language understanding looks like.

Jon Hernandez

77,663 次观看 • 5 个月前

Yann LeCun: I'm not interested in LLMs anymore - they're the past. The future is in four more interesting areas: machines that understand the physical world, persistent memory, reasoning, and planning.

Yann LeCun: I'm not interested in LLMs anymore - they're the past. The future is in four more interesting areas: machines that understand the physical world, persistent memory, reasoning, and planning.

Victor

525,463 次观看 • 1 年前

Yann LeCun says LLMs are strongest in domains where language itself is the substrate of reasoning, like math and code They can solve problems, prove theorems, and write programs — but they are not creative mathematicians, software architects, or computer scientists "their role is to help humans build"

Yann LeCun says LLMs are strongest in domains where language itself is the substrate of reasoning, like math and code They can solve problems, prove theorems, and write programs — but they are not creative mathematicians, software architects, or computer scientists "their role is to help humans build"

Haider.

346,753 次观看 • 28 天前

Yann LeCun says despite LLMs passing bar exams, fundamental inventions remain missing. We still lack domestic robots, fully autonomous cars, and systems with true physical understanding and persistent memory. "your house cat is way smarter than the biggest LLMs"

Yann LeCun says despite LLMs passing bar exams, fundamental inventions remain missing. We still lack domestic robots, fully autonomous cars, and systems with true physical understanding and persistent memory. "your house cat is way smarter than the biggest LLMs"

Haider.

37,721 次观看 • 1 年前

📁 Yann LeCun says that scaling models will not get us to human intelligence. He explains that the industry remains obsessed with making LLMs bigger, but that this path is fundamentally broken. It does not matter how many parameters we add or how many clusters we build, because the models only imitate language patterns. Human intelligence does not emerge from size, it emerges from understanding the world.

📁 Yann LeCun says that scaling models will not get us to human intelligence. He explains that the industry remains obsessed with making LLMs bigger, but that this path is fundamentally broken. It does not matter how many parameters we add or how many clusters we build, because the models only imitate language patterns. Human intelligence does not emerge from size, it emerges from understanding the world.

Jon Hernandez

154,597 次观看 • 6 个月前

Yann LeCun says the real world is far more complex than the world of language LLMs can accumulate knowledge, but they fail with high-dimensional, continuous, noisy sensory data "the next revolution is physical AI" Systems that can truly plan, reason, and understand the physical environment

Yann LeCun says the real world is far more complex than the world of language LLMs can accumulate knowledge, but they fail with high-dimensional, continuous, noisy sensory data "the next revolution is physical AI" Systems that can truly plan, reason, and understand the physical environment

Haider.

66,616 次观看 • 4 个月前

Yann LeCun (Yann LeCun ) beautifully explains how the architecture and principles used to train LLMs can not be extended to teach AI the real-world intelligence. In 1 line: LLMs excel where intelligence equals sequence prediction over symbols. Real-world intelligence requires learned world models, abstraction, causality, and action planning under uncertainty, which current next-token training does not provide. He says current LLMs learn by predicting the next token. That objective works very well when the task itself can be reduced to manipulating discrete symbols and sequences. Math, physics problem solving on paper, and coding fit this pattern because success largely comes from searching and composing the right sequences of symbols, equations, or program tokens. With enough data and scale, these models get very good at that kind of structured sequence prediction. Real-world intelligence is different. The physical world is continuous, noisy, uncertain, and high dimensional. To act in it, a system needs internal models that capture objects, dynamics, causality, constraints from the body, and the outcomes of actions over time. Humans and animals build abstract representations from rich sensory streams, then make predictions in that abstract space, not at the raw pixel level. That is why a child can learn intuitive physics, plan multi-step actions, and adapt quickly in new situations with little data. His claim about saturation follows from this gap. Scaling token prediction keeps improving symbol manipulation tasks like math and code, but it hits limits on embodied reasoning and common sense because text alone does not provide the right learning signals for world models. Predicting the next word cannot efficiently teach contact forces, affordances, occlusion, friction, or how actions change the state of the environment. For that, he argues we need architectures that learn abstractions from sensory data and predict futures in abstract latent spaces, then use those predictions to plan actions toward goals with built-in guardrails. --- From 'Pioneer Works' YT Channel (link in comment)

Yann LeCun (Yann LeCun ) beautifully explains how the architecture and principles used to train LLMs can not be extended to teach AI the real-world intelligence. In 1 line: LLMs excel where intelligence equals sequence prediction over symbols. Real-world intelligence requires learned world models, abstraction, causality, and action planning under uncertainty, which current next-token training does not provide. He says current LLMs learn by predicting the next token. That objective works very well when the task itself can be reduced to manipulating discrete symbols and sequences. Math, physics problem solving on paper, and coding fit this pattern because success largely comes from searching and composing the right sequences of symbols, equations, or program tokens. With enough data and scale, these models get very good at that kind of structured sequence prediction. Real-world intelligence is different. The physical world is continuous, noisy, uncertain, and high dimensional. To act in it, a system needs internal models that capture objects, dynamics, causality, constraints from the body, and the outcomes of actions over time. Humans and animals build abstract representations from rich sensory streams, then make predictions in that abstract space, not at the raw pixel level. That is why a child can learn intuitive physics, plan multi-step actions, and adapt quickly in new situations with little data. His claim about saturation follows from this gap. Scaling token prediction keeps improving symbol manipulation tasks like math and code, but it hits limits on embodied reasoning and common sense because text alone does not provide the right learning signals for world models. Predicting the next word cannot efficiently teach contact forces, affordances, occlusion, friction, or how actions change the state of the environment. For that, he argues we need architectures that learn abstractions from sensory data and predict futures in abstract latent spaces, then use those predictions to plan actions toward goals with built-in guardrails. --- From 'Pioneer Works' YT Channel (link in comment)

Rohan Paul

104,460 次观看 • 5 个月前

Ilya on LLMs understanding the world: "predicting the next token well, means that you understand the underlying reality that let to the creation of that token" Seem like the opposite view of Yann.

Ilya on LLMs understanding the world: "predicting the next token well, means that you understand the underlying reality that let to the creation of that token" Seem like the opposite view of Yann.

Lior Alexander

1,222,003 次观看 • 2 年前

📁 Yann LeCun, Chief AI Scientist at Meta, says many people misunderstand what a world model really is. He explains that it is not about reproducing the world in full detail like a visual simulator. He warns that generating video may look impressive, but it does not guarantee the system has learned the underlying dynamics of reality.

📁 Yann LeCun, Chief AI Scientist at Meta, says many people misunderstand what a world model really is. He explains that it is not about reproducing the world in full detail like a visual simulator. He warns that generating video may look impressive, but it does not guarantee the system has learned the underlying dynamics of reality.

Jon Hernandez

36,093 次观看 • 5 个月前

Yann LeCun says robotics could become a huge market, but the industry still doesn't know how to make humanoid robots useful The missing piece is AI that can understand the physical world and handle tasks with generality and adaptability "LLMs are useful, but for different problems"

Yann LeCun says robotics could become a huge market, but the industry still doesn't know how to make humanoid robots useful The missing piece is AI that can understand the physical world and handle tasks with generality and adaptability "LLMs are useful, but for different problems"

Haider.

13,469 次观看 • 19 天前

Yann LeCun (Yann LeCun ) explains why LLMs are so limited in terms of real-world intelligence. Says the biggest LLM is trained on about 30 trillion words, which is roughly 10 to the power 14 bytes of text. That sounds huge, but a 4 year old who has been awake about 16,000 hours has also taken in about 10 to the power 14 bytes through the eyes alone. So a small child has already seen as much raw data as the largest LLM has read. But the child’s data is visual, continuous, noisy, and tied to actions: gravity, objects falling, hands grabbing, people moving, cause and effect. From this, the child builds an internal “world model” and intuitive physics, and can learn new tasks like loading a dishwasher from a handful of demonstrations. LLMs only see disconnected text and are trained just to predict the next token. So they get very good at symbol patterns, exams, and code, but they lack grounded physical understanding, real common sense, and efficient learning from a few messy real-world experiences. --- From 'Pioneer Works' YT channel (link in comment)

Yann LeCun (Yann LeCun ) explains why LLMs are so limited in terms of real-world intelligence. Says the biggest LLM is trained on about 30 trillion words, which is roughly 10 to the power 14 bytes of text. That sounds huge, but a 4 year old who has been awake about 16,000 hours has also taken in about 10 to the power 14 bytes through the eyes alone. So a small child has already seen as much raw data as the largest LLM has read. But the child’s data is visual, continuous, noisy, and tied to actions: gravity, objects falling, hands grabbing, people moving, cause and effect. From this, the child builds an internal “world model” and intuitive physics, and can learn new tasks like loading a dishwasher from a handful of demonstrations. LLMs only see disconnected text and are trained just to predict the next token. So they get very good at symbol patterns, exams, and code, but they lack grounded physical understanding, real common sense, and efficient learning from a few messy real-world experiences. --- From 'Pioneer Works' YT channel (link in comment)

Rohan Paul

638,969 次观看 • 3 个月前

📁 Yann LeCun, Chief AI Scientist at Meta, says language is not the peak of intelligence, it is the easy part. Predicting the next word is simple because language is made of finite symbols. The real world is continuous, noisy and chaotic, and even a cat navigates it better than our best models. True intelligence begins where text ends.

📁 Yann LeCun, Chief AI Scientist at Meta, says language is not the peak of intelligence, it is the easy part. Predicting the next word is simple because language is made of finite symbols. The real world is continuous, noisy and chaotic, and even a cat navigates it better than our best models. True intelligence begins where text ends.

Jon Hernandez

56,133 次观看 • 3 个月前

📁 Yann LeCun says language models do extract meaning, but only at a superficial level. Unlike humans, their intelligence is not grounded in physical reality or common sense. They answer many questions well, but break down when faced with new situations because they do not truly understand the world they describe.

📁 Yann LeCun says language models do extract meaning, but only at a superficial level. Unlike humans, their intelligence is not grounded in physical reality or common sense. They answer many questions well, but break down when faced with new situations because they do not truly understand the world they describe.

Jon Hernandez

49,318 次观看 • 5 个月前

📁 Demis Hassabis says that one of the main reasons we are not yet at AGI is the lack of consistency. He explains that current systems can solve International Math Olympiad level problems but still fail at basic logic tasks or even chess. They are uneven intelligences, exceptional in some areas and surprisingly weak in others, which falls short of true AGI.

📁 Demis Hassabis says that one of the main reasons we are not yet at AGI is the lack of consistency. He explains that current systems can solve International Math Olympiad level problems but still fail at basic logic tasks or even chess. They are uneven intelligences, exceptional in some areas and surprisingly weak in others, which falls short of true AGI.

Jon Hernandez

10,541 次观看 • 5 个月前

📁 Yann LeCun, Meta’s Chief AI Scientist and Turing Award winner, says today’s AI only looks intelligent. It is very good at manipulating language. But it still cannot understand the physical world, maintain real memory or truly plan. The next generation of AI will need to understand the world, not just talk about it.

📁 Yann LeCun, Meta’s Chief AI Scientist and Turing Award winner, says today’s AI only looks intelligent. It is very good at manipulating language. But it still cannot understand the physical world, maintain real memory or truly plan. The next generation of AI will need to understand the world, not just talk about it.

Jon Hernandez

22,491 次观看 • 3 个月前

📁 Yann LeCun, an expert in artificial intelligence and a pioneer of deep learning, explains that current systems do not understand the physical world and lack real memory or reasoning. He adds that, despite the vast resources spent on training and refining them, they remain far behind the learning ability of a four-year-old child.

📁 Yann LeCun, an expert in artificial intelligence and a pioneer of deep learning, explains that current systems do not understand the physical world and lack real memory or reasoning. He adds that, despite the vast resources spent on training and refining them, they remain far behind the learning ability of a four-year-old child.

Jon Hernandez

24,677 次观看 • 7 个月前

📁 Yann LeCun says that AGI does not exist and is a poorly defined concept built around an idealized view of human intelligence. He argues that humans are highly specialized, good at navigating the physical and social world, but weak at many other tasks. We seem general only because we can imagine the problems we are good at. For LeCun, AGI mistakes human limits for universal laws of intelligence.

📁 Yann LeCun says that AGI does not exist and is a poorly defined concept built around an idealized view of human intelligence. He argues that humans are highly specialized, good at navigating the physical and social world, but weak at many other tasks. We seem general only because we can imagine the problems we are good at. For LeCun, AGI mistakes human limits for universal laws of intelligence.

Jon Hernandez

28,104 次观看 • 5 个月前

Yann LeCun says LLMs are not a bubble in value or investment; they will power many useful apps and justify big infra The bubble is believing LLMs alone will reach human-level intelligence Progress needs breakthroughs, not just more data/compute "we're missing something big"

Yann LeCun says LLMs are not a bubble in value or investment; they will power many useful apps and justify big infra The bubble is believing LLMs alone will reach human-level intelligence Progress needs breakthroughs, not just more data/compute "we're missing something big"

Haider.

321,406 次观看 • 7 个月前

Yann LeCun argues that large language models (LLMs) cannot reach human-level or superintelligence just by scaling. He says the current LLM paradigm is hitting its limits. Many researchers are now exploring “agentic systems,” but building them on top of LLMs alone is flawed. LLMs can't plan actions well because they don’t truly understand or predict consequences. To get intelligent behavior, we need something fundamentally different.

Yann LeCun argues that large language models (LLMs) cannot reach human-level or superintelligence just by scaling. He says the current LLM paradigm is hitting its limits. Many researchers are now exploring “agentic systems,” but building them on top of LLMs alone is flawed. LLMs can't plan actions well because they don’t truly understand or predict consequences. To get intelligent behavior, we need something fundamentally different.

Wes Roth

71,792 次观看 • 4 个月前