Yann LeCun holds a pen upright on its tip and asks a simple question: what happens when you let go? Even a toddler knows the pen will topple over — but no human would bother guessing which way it falls. A large language model, he argues, would try to generate a single prediction based on statistical patterns, and it would almost certainly be wrong.
"They basically just accumulate knowledge... They can regurgitate something, you train them to regurgitate, but they're not particularly smart. They don't have an underlying understanding," says LeCun, one of the leading figures in artificial intelligence.
“Yann LeCun says current AI is 'not smart', has raised $1bn for a new system called JEPA.”
After a decade as chief AI scientist at Facebook-owner Meta, LeCun left in 2025 to found Advanced Machine Intelligence Labs (AMI Labs) in Paris. His goal is to move beyond systems like ChatGPT, Claude and Gemini which, he says, will never handle complicated real-world situations — like getting a robot to do household chores.
"We don't have robots that are nearly as good at understanding the physical world as a rat," he told the BBC on the sidelines of VivaTech, France's leading technology conference.
"They're not a path towards human level or human-like intelligence, or even animal-like intelligence, because they cannot deal with real world data, they just are not built for that."
So AMI Labs is developing a new type of artificial intelligence not based on the tech behind ChatGPT. The system, called Joint Embedding Predictive Architecture (JEPA), creates abstractions of the real world that allow it to assess the outcomes of actions. These abstractions involve difficult maths but essentially filter out useless information.
Investors see potential. Earlier this year AMI Labs announced it had raised more than $1bn (£760m) in seed funding — one of the biggest such rounds in Europe. Backers include US computer chip giant Nvidia and the fund that manages the private wealth of Amazon-founder Jeff Bezos.
Large language models are extremely good at some things — coding, mathematical problems, generating text — but LeCun says these are well-defined, predictable problems. In the real world, a bewildering array of outcomes to any action requires a more flexible intelligence.
What comes next may be a system that doesn't just regurgitate knowledge but reasons about physical reality. Whether JEPA can deliver remains to be seen, but one of AI's most respected figures is betting $1bn that the answer lies beyond the current hype.