After 12 years at Meta, Yann LeCun, the company's Chief AI Scientist and a prominent figure in the field of artificial intelligence, is departing to launch his own independent AI research venture. LeCun, a Turing Award winner, confirmed his long-rumored departure via a LinkedIn post. He joined Meta in 2013 and went on to co-found FAIR (Facebook AI Research), which became a highly influential AI research group. He stepped down as FAIR's director in 2018, but continued his work as Meta's Chief AI Scientist.
LeCun's new venture will focus on "Advanced Machine Intelligence (AMI)", with the goal of developing AI systems capable of understanding the physical world, reasoning, and planning complex actions. He envisions AI that goes beyond current generative models. This involves creating systems with persistent memory, real-world understanding, and long-horizon reasoning. According to LeCun, this is essential for "the next big revolution in AI".
Several factors contributed to LeCun's decision. Meta has been reorganizing its AI teams to focus more on product-driven and commercial AI efforts. The creation of a new "Superintelligence Labs" unit, led by Alexandr Wang, resulted in LeCun reporting to Wang, a shift that, coupled with layoffs affecting FAIR staff, signaled a change in the company's research priorities. LeCun has also voiced skepticism about large language models (LLMs), arguing that they are insufficient for achieving truly intelligent systems. He advocates for approaches that enable machines to reason, plan, and understand the physical world. His recent research has focused on "world modeling"—developing AI systems capable of quickly learning about the physical world. He has been critical of relying solely on text and large language models.
Despite his departure, Meta will remain a partner in LeCun's new company. LeCun expressed gratitude to Mark Zuckerberg and other Meta executives for their support of FAIR and the AMI program. He believes that pursuing AMI in an independent entity will maximize its broad impact across various sectors, some overlapping with Meta's interests and others operating independently. LeCun will remain at Meta until the end of the year.
LeCun's departure underscores the challenges tech firms face in balancing visionary research with commercial pressures in the rapidly evolving AI landscape. His move could catalyze a broader talent shift, inspiring other researchers to pursue independent AI innovation. Industry experts are closely watching the development, as LeCun's new venture could significantly impact the future of AI. His work at FAIR has already produced innovations like V-JEPA-2, a model trained on videos to model physical interactions. His new startup is expected to continue this line of research. LeCun's new technology may apply to industries in and out of Meta's commercial interests.
