Artificial intelligence systems such as ChatGPT lack the intelligence of a rat and will never achieve human-like understanding, one of the field's leading figures has claimed – as the Bank of England warned that AI poses a risk to the UK's financial stability.
Yann LeCun, who left his role as chief AI scientist at Meta in 2025 to found Advanced Machine Intelligence Labs (AMI Labs) in Paris, told the VivaTech conference that current AI is not a path towards human or animal-level intelligence because it cannot deal with real-world data. "We don't have robots that are nearly as good at understanding the physical world as a rat," he said.
“AI is 'not smart', says Yann LeCun, as Bank of England warns of financial risk from AI.”
LeCun's company is developing a new type of AI called Joint Embedding Predictive Architecture (JEPA), which creates abstractions of the real world to assess outcomes. The system is designed to handle the unpredictability of physical reality – something that Large Language Models (LLMs) like ChatGPT cannot do, as they rely on statistical patterns and often generate wrong predictions. To illustrate, LeCun held up a pen upright: a toddler would know it would fall, but an LLM might guess the direction based on statistics and almost certainly be wrong.
"They [LLMs] 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," LeCun said. He argued that LLMs are good at well-defined tasks like coding and mathematics, but fail at the bewildering array of outcomes in the real world.
Investors have backed LeCun's approach. AMI Labs announced earlier this year that it had raised more than $1bn (£760m) in seed funding, one of the largest such rounds in Europe, with backing from Nvidia and the fund that manages the private wealth of Amazon-founder Jeff Bezos.
Separately, the Bank of England has said the UK financial system is at increased risk from artificial intelligence. LeCun's critique of current AI systems highlights the potential dangers of deploying technology that lacks robust understanding of real-world dynamics in critical sectors like finance.
Whether JEPA can overcome these limitations remains to be seen, but LeCun is convinced a fundamental shift is needed – and investors have bet more than $1bn on it.
