The head of artificial intelligence at Meta, Yann LeCun, would be preparing to leave the company to found his own startup, according to inform Financial Times. The departure of the prestigious researcher, winner of the Turing Award and considered one of the fathers of modern AI, symbolizes the radical change that Mark Zuckerberg is giving to Meta’s strategy around AI.
The changing of the guard. LeCun, who led the Fundamental AI Research Laboratory (FAIR) since 2013, is now in an uncomfortable position within Meta. This summer, Zuckerberg hired Alexander Wang28, to lead a new “superintelligence” team, paying $14.3 billion to take 49% of Scale AI, the data labeling startup Wang had founded. As a result of this restructuring, LeCun went from reporting to chief product officer Chris Cox to reporting to Wang, according to account Financial Times.
A philosophical divorce. The tension is not only organizational, but conceptual. LeCun has long publicly defended that the language models on which Zuckerberg has focused his strategy are “useful” but will never be able to reason or plan like humans. His bet from FAIR has been different: the so-called “world models”AI systems that learn from the physical environment through videos and spatial data, not just language. A path that, according to LeCun himself, could take a decade to bear fruit.
Meta’s problems with AI. Zuckerberg’s reorganization comes after several setbacks. The launch of Call 4 It has not gone as the company would have liked, falling below the most advanced proposals from OpenAI, Google and Anthropic. Additionally, Meta AI, the company’s chatbot, has also not gained traction among users.
Meanwhile, Zuckerberg has hired dozens of engineers and competing researchers with pay packages of up to $100 million, creating a dedicated team called TBD Lab to accelerate the development of new versions of its language models.
The cost of pivoting. The shift toward practical AI appears to have generated internal chaos. Sources cited by TechCrunch In August they revealed the frustration of new hires when facing the bureaucracy of a large company, while the previous generative AI team saw its scope reduced.
In October, Meta laid off 600 people of its AI research unit to cut costs and accelerate product launches. Also in May Joelle Pineau left the companyvice president of AI research, who joined Canadian startup Cohere.
What’s coming now. According to two sources Cited by the Financial Times, LeCun’s new project will focus on continuing his work on world models, and he has already started talks to raise funding. His departure, scheduled for the coming months, represents more than the departure of a brilliant scientist: it is confirmation that Meta’s old long-term focus has been relegated by the urgency of competing in the short term with more practical solutions.
As Wall Street pressures Zuckerberg to justify an investment in AI that could exceed $100 billion In 2025, the company would be losing one of its most recognized brains along the way.
Cover image | Goal and AFP
In Xataka | AI was supposed to reduce costs and reduce staff. The Coca-Cola ad illustrates how much we were wrong
GIPHY App Key not set. Please check settings