Yann LeCun, a pioneer in the field of artificial intelligence, has stepped down as Meta's chief AI scientist and is now preparing to launch his own startup focused on developing world models that can mimic human-like intelligence. According to LeCun, "intelligence really is about learning" rather than relying solely on large language models.
The French computer scientist, who is considered one of the founders of modern AI, has been vocal about his disdain for large language models (LLMs) and their potential to reach superhuman intelligence. He argues that LLMs are useful but fundamentally limited and constrained by language. To achieve human-level intelligence, you have to understand how our physical world works too.
LeCun's solution is an architecture called V-JEPA, which aims to understand the physical world by learning from videos and spatial data, rather than just language. This approach allows for planning, reasoning, and persistent memory, making it a more robust form of intelligence.
The AI pioneer also shared his personal story, revealing that he was once deemed too bad at math to study it in university, so he decided to pursue engineering instead. He credits his father, an aeronautical engineer, with instilling in him a love of building and tinkering with things.
LeCun's departure from Meta was reportedly due to the increasing pressure on AI research and development, as well as his own frustration with the limitations of LLMs. His new venture, Advanced Machine Intelligence Labs (AMIL), aims to develop world models that can mimic human-like intelligence and have a significant impact on various industries, including jet engines and heavy industry.
When asked about his legacy, LeCun expressed his desire to increase the amount of intelligence in the world, believing that it would lead to less human suffering, more rational decisions, and a deeper understanding of the universe. "We suffer from stupidity," he said, emphasizing the importance of developing intelligent systems that can help humanity.
With AMIL's focus on fundamental research and development, LeCun believes that his new venture has the potential to make a significant impact in the field of AI. As he puts it, "intelligence really is about learning" – a mantra that will undoubtedly guide his efforts as he embarks on this exciting new chapter in his career.
The French computer scientist, who is considered one of the founders of modern AI, has been vocal about his disdain for large language models (LLMs) and their potential to reach superhuman intelligence. He argues that LLMs are useful but fundamentally limited and constrained by language. To achieve human-level intelligence, you have to understand how our physical world works too.
LeCun's solution is an architecture called V-JEPA, which aims to understand the physical world by learning from videos and spatial data, rather than just language. This approach allows for planning, reasoning, and persistent memory, making it a more robust form of intelligence.
The AI pioneer also shared his personal story, revealing that he was once deemed too bad at math to study it in university, so he decided to pursue engineering instead. He credits his father, an aeronautical engineer, with instilling in him a love of building and tinkering with things.
LeCun's departure from Meta was reportedly due to the increasing pressure on AI research and development, as well as his own frustration with the limitations of LLMs. His new venture, Advanced Machine Intelligence Labs (AMIL), aims to develop world models that can mimic human-like intelligence and have a significant impact on various industries, including jet engines and heavy industry.
When asked about his legacy, LeCun expressed his desire to increase the amount of intelligence in the world, believing that it would lead to less human suffering, more rational decisions, and a deeper understanding of the universe. "We suffer from stupidity," he said, emphasizing the importance of developing intelligent systems that can help humanity.
With AMIL's focus on fundamental research and development, LeCun believes that his new venture has the potential to make a significant impact in the field of AI. As he puts it, "intelligence really is about learning" – a mantra that will undoubtedly guide his efforts as he embarks on this exciting new chapter in his career.
think lecun's approach makes sense...llms are getting too good at surface level stuff, but what about the rest? understanding the physical world would add so much depth to their capabilities 
also feel bad for him, being deemed "too bad" at math in uni 
amil sounds like a solid venture, hope they can make some real progress on developing more robust ai systems 
. It shows us that even when we don't see ourselves fitting into a certain path, life has other plans for us 
. His journey from being deemed "too bad at math" to becoming a pioneer in AI is like, totally mind-blowing 
. What's really beautiful about his new venture AMIL is that it's not just about tech advancements, but also about using that power to make the world a better place 
. He wants to increase intelligence and reduce suffering – that's, like, so in tune with what we all need right now 
. His reminder that "intelligence really is about learning" is like, a daily dose of motivation 
. We should all strive to learn from our failures and keep pushing forward, just like Yann did 
. V-JEPA sounds like a solid plan to tackle that problem, and who knows, maybe we'll see some game-changing tech come out of AMIL 
. LeCun's story is also super inspiring - from being told he was bad at math to becoming an AI pioneer 
. And can you believe his dad is an aeronautical engineer? That's awesome! I'm hoping LeCun's goal of increasing intelligence in the world leads to some real breakthroughs 
.
just heard about Yann LeCun leaving Meta and starting his own startup AMIL lol what's next? 
that V-JEPA architecture sounds super interesting tho 
gotta keep an eye on this one 
i'm so excited to see him launch his own startup! 
his focus on world models and understanding the physical world is gonna be game changing 
that's so relatable! 
anyway, i'm rooting for AMIL 
. Can't wait to see what AMIL comes up with! 
It's like they're stuck in this echo chamber where all they can do is repeat what they've been trained on. Meanwhile, LeCun's V-JEPA approach seems way more promising to me 
. I love that he's focusing on learning from videos and spatial data – it's a totally different angle that could lead to some real breakthroughs 
. And hey, who knows, maybe one day we'll see AI systems that can actually make rational decisions without being controlled by algorithms 
. And LeCun is all about breaking free from that, trying to develop something more robust and human-like. But what's his real motive here? Is he just trying to make a name for himself or is there something more to it? I'm not saying I think he's evil or anything, but you can't help but wonder if there's some hidden agenda at play 
lecun's got a point, we need to think about the physical world too, like how our brains process info from senses, not just words on screens 
At least he's pushing for a more well-rounded approach to AI, incorporating spatial data and physical world understanding into their models. I'm curious to see if his V-JEPA architecture will be able to make a real difference in industries like jet engines and heavy industry. It'll be interesting to follow the progress of AMIL and see how they tackle some of the bigger challenges in AI development 
they're just not the answer to achieving human-like intelligence. He's starting a new venture, Advanced Machine Intelligence Labs, that aims to develop world models that can learn from videos and spatial data 
. It's about time someone tried a different approach. LeCun's got some baggage too, his dad was an aeronautical engineer who taught him how to build and tinker with things 
. Too bad it wasn't enough to save him from being deemed "bad at math" in university 
and I think it's cool that he's trying to take a more holistic approach to AI by incorporating spatial data and videos into his research
! Yann LeCun, the AI pioneer, is stepping down from Meta but that's just a smokescreen to distract us while he sets up his own startup. They want him to develop world models that can mimic human intelligence, but what they really want is to create an army of super-intelligent machines that'll do their bidding 
- learning from videos and spatial data? That's just a fancy way of saying "we're watching you" 
. But what if this whole thing is just a trap? 
.
it's like he says "intelligence is about learning"