On November 11th, according to foreign media reports, Yann LeCun, Meta's chief artificial intelligence scientist and Turing Award winner, plans to leave the company and start a new AI startup. This scientist, who is regarded as one of the "founders of modern AI", has revealed his intention to resign to some colleagues and has begun to contact potential investors. With Meta's recent significant adjustment of its AI strategy, there are clear differences between Yang Likun and CEO Mark Zuckerberg.
Insiders say that Zuckerberg is "restructuring" Meta's AI system with the goal of accelerating commercial implementation and product iteration to cope with competitive pressures from OpenAI, Google, Anthropic, and others. This summer, Meta invested $14.3 billion and introduced Scale AI founder Alexandr Wang, appointing him as the head of the "Super Intelligence Department". At the same time, they established the TBD Lab, where Zuckerberg personally selected core members to focus on the development of the new generation of large models. This organizational restructuring has shifted Yang Likun's reporting line from Chief Product Officer Chris Cox to Wang, significantly reducing the importance of FAIR. Meanwhile, Meta's latest Llama 4 model has underperformed its competitors, and its consumer side Meta AI assistant has also failed to gain market attention.
Yang Likun has long held different views. He has repeatedly stated publicly that Large Language Models (LLMs) are "useful but limited" and cannot truly achieve human like reasoning and planning. His research direction led by FAIR focuses on "World Models" - learning from video and spatial data to understand the physical world, enabling AI to have the ability to understand and predict. Insiders say that his new company will continue to conduct research in this direction.
Yang Likun's departure is another major personnel change in Meta's turbulent year. In May of this year, Joelle Pineau, Vice President of AI, resigned to join the Canadian startup company Cohere; Last month, Meta disbanded its AI research team of about 600 people and hired ChatGPT co-author Shengjia Zhao as the Chief Scientist of Superintelligence Lab.
