Meta’s chief AI scientist Yann LeCun is leaving Big Tech for a new, riskier bet. And his former company, which will soon be established, will be a partner, but not an investor.
Last month, LeCun announced on LinkedIn that he was leaving Meta after 12 years. He spent five of those years as the founding director of Fundamental AI Research (FAIR), formerly Meta’s primary AI research institute, and seven years as the company’s chief AI scientist.
However, this year, Meta reorganized its AI division around the Superintelligence Institute, which is headed by Scale AI founder Alexander Wang.
“I am forming a startup to continue the Advanced Machine Intelligence Research Program (AMI) that I have pursued over the past several years with colleagues from FAIR, New York University, and elsewhere,” LeCun wrote on LinkedIn in November. “The startup’s goal is to bring about the next big revolution in AI: systems that understand the physical world, have persistent memory, can reason, and can plan complex courses of action.”
However, he does not completely sever ties with Meta. Speaking at the AI-Pulse event in Paris on Thursday, LeCun said Mehta would not be an investor, but would be a partner in the new effort.
“This new architecture is a project that Mark Zuckerberg really likes. He thinks it’s probably the future,” LeCun said of his ambition to build world models, or advanced machine learning systems, that help humans make decisions and predictions from abstract representations of the world. This is different from LLM, which powers the most famous AI chatbots, which relies solely on language.
These ambitions are beyond Mr. Mehta’s scope, he said.
“He and I both realized that the potential applications of this were a little bit beyond what Meta was interested in,” he said, referring to Zuckerberg.
“The number of applications is so wide-ranging that it would have been better to do it as an independent organization.”
Meta’s AI ambitions are shifting toward superintelligence, a still-theoretical AI that can reason as well or better than humans.
The company established a super intelligence lab in June. Zuckerberg said he hopes this will help the company build a “personal superintelligence,” a term he uses to describe AI systems that could eventually surpass human capabilities.
Neither Meta nor LeCun responded to requests for comment from Business Insider.
