After 12 years with Meta, Chief AI Scientist Yann LeCun officially announced his departure on November 20th to found his own AI startup. This pioneer in deep learning led Meta's FAIR lab from 2013, eventually rising to Chief AI Scientist, during which time he laid a crucial foundation for Meta's core technologies in computer vision and fundamental models.
Yan stated that his new company will continue his Advanced Machine Intelligence (AMI) research program with FAIR and New York University, while maintaining collaboration with Meta. Its goal is to advance AI systems to achieve real-world understanding, long-term memory, multi-step reasoning, and planning capabilities, breaking through current technological limitations. He emphasized that an independent company allows for greater freedom to disseminate research findings, avoiding the constraints of short-term commercial pressures imposed by large corporations.
Notably, LeCun has consistently been critical of large language models, publicly opposing the idea that "human-level AI can be achieved solely through scaling," and even advising young researchers to "stay away from large language models." According to Bloomberg, this stance differs from Meta's current strategy of focusing on productization models. The internal competition for resources, with long-term exploratory research like AMI gradually giving way to short-term projects targeting competitors like OpenAI and Google, may have been a key factor in his departure.
