Meta's top AI scientist Yann LeCun is making the jump from the company he helped build its artificial intelligence capabilities at to start his own venture focused on developing physical AI systems.
The move comes as Meta doubles down on advanced AI research, including superintelligence, following a major restructuring. LeCun was once instrumental in launching the company's Fundamental AI Research (FAIR) lab, but his work is now taking a backseat as he focuses on "world models", which are designed to understand the physical world rather than just generating language.
The French-American computer scientist has long been critical of large language models, or LLMs, describing them as a "dead end" for reaching human-like AI. Instead, LeCun believes that systems capable of perceiving their environment and grasping physical concepts like gravity are key to developing true intelligence.
He is not alone in this pursuit. Other researchers and companies, such as Stanford's Fei-Fei Li and Google DeepMind, are exploring world models, which aim to give AI "spatial intelligence". Nvidia is also pushing into physical AI with its Cosmos products.
LeCun's departure from Meta marks another significant shake-up for the company, following the loss of his former boss Joelle Pineau in April. While LeCun did not comment on his reasons for leaving, it's clear that he believes world models hold the key to developing true intelligence and surpassing human capabilities.
The move comes as Meta doubles down on advanced AI research, including superintelligence, following a major restructuring. LeCun was once instrumental in launching the company's Fundamental AI Research (FAIR) lab, but his work is now taking a backseat as he focuses on "world models", which are designed to understand the physical world rather than just generating language.
The French-American computer scientist has long been critical of large language models, or LLMs, describing them as a "dead end" for reaching human-like AI. Instead, LeCun believes that systems capable of perceiving their environment and grasping physical concepts like gravity are key to developing true intelligence.
He is not alone in this pursuit. Other researchers and companies, such as Stanford's Fei-Fei Li and Google DeepMind, are exploring world models, which aim to give AI "spatial intelligence". Nvidia is also pushing into physical AI with its Cosmos products.
LeCun's departure from Meta marks another significant shake-up for the company, following the loss of his former boss Joelle Pineau in April. While LeCun did not comment on his reasons for leaving, it's clear that he believes world models hold the key to developing true intelligence and surpassing human capabilities.
I've been following his work for a while now, and it's awesome to see him taking on this new challenge. He's always been ahead of the curve when it comes to AI research, and I think his focus on world models is exactly what we need to move forward. It's like he's saying, "Hey, let's not just talk to each other, but actually experience the world!" 
As a parent, I'm all about encouraging kids to explore, discover, and learn through their senses, so this whole concept resonates with me. Plus, think of all the amazing things we could achieve if AI can truly understand our physical environment! It's like having a superpower! 
That would be wild!
i think its cool that yann lecun is starting his own thing now 
, imo meta should focus more on physical ai like him instead of just churning out more lllms 
, and honestly its about time someone spoke up against large language models 
, the way they're being used now seems kinda sketchy to me 
I'm low-key hyped for Yann LeCun's new venture! He's been super vocal about how large language models are just a starting point for AI, and I think his focus on physical AI systems is so important 
. We need to get beyond just chatty bots and create systems that can actually interact with the world in a meaningful way 
. LeCun's got some serious clout in this space, and I'm excited to see what he comes up with next 
... Yann LeCun making the jump from Meta to start his own venture is like, totally expected 
. I mean, world models are where it's at! 
. And let's be real, who wouldn't want to work on something that could give AI spatial intelligence and make it truly intelligent? 
...
. If he can make these physical AI systems work, we'll be talking about robots that can actually navigate and understand their surroundings 
? And what about the timeline for all this? Like, when can we expect to see these world models in action? 
. But, at the same time, it's hard not to wonder what the implications are 
.
but seriously yann lecun sounds like a genius i wish i could build something like his world models idk how he does it 
what if we can make robots that can actually help us in everyday life like cooking and cleaning? 
now im not surprised tho lol lecun has always been like "lmms r a dead end" 
and ppl like him who think we need to focus on physical ai are finally getting the recognition they deserve 
in AI research, and I'm so stoked he's starting his own thing 
. It's time to ditch those language-only LLMs 
and get AI to learn from the physical world 
. Let's hope Yann's departure from Meta doesn't mean we'll see him for a while 
, because I need more of his genius-level thinking 
Yann LeCun's move to start his own venture focused on physical AI systems is a huge leap for humanity! It shows how some people are willing to take risks and challenge the status quo. The fact that other top researchers like Fei-Fei Li and Google DeepMind are also working on world models gives me hope that we might see some groundbreaking advancements in AI soon! 
. Still, kudos to him for taking a chance on something new 
, so it's like he's taking his own advice. World models are definitely the way to go, but have you seen those AI robots in "Westworld" 
. yann lecun's been saying that large language models are a dead end for reaching true intelligence and now he's starting his own thing focused on physical AI systems 
.