Yann LeCun, one of the three leading scholars of artificial intelligence (AI), a New York University professor and Meta AI chief scientist, is reportedly leaving Meta to found a Start - Up.
The Financial Times (FT), citing sources, reported on the 11th (local time) that LeCun recently told acquaintances he plans to leave Meta, Facebook's parent company. Since 2013, LeCun has served as a vice president with the title of chief AI scientist at Facebook, leading the Fundamental AI Research (FAIR) lab.
LeCun's decision comes as Meta Chief Executive Officer (CEO) Mark Zuckerberg is moving to overhaul the AI organization, drawing attention. Zuckerberg is pushing a strategy to move away from the long-term AI research LeCun has led and release Generative AI models and products based on large language models (LLM) more quickly. With evaluations that Meta has fallen behind rivals like Google and OpenAI, the company has entered a speed race.
To that end, Meta has set a goal to be the first to develop "superintelligence," going beyond artificial general intelligence (AGI) with human-level intelligence. In June, Meta invested $14.3 billion in the promising AI Start - Up Scale AI and hired 28-year-old founder Alexandr Wang, known as a "genius developer," as chief AI officer (CAIO) to lead the Meta Superintelligence Lab (MSL).
It also laid off hundreds of researchers and, in August, carried out a reorganization dividing the AI organization into four teams: the "TBD Lab," which oversees LLMs, FAIR, the product and applied research team, and the infrastructure team. As a result, LeCun has been reporting to Wang, the CAIO.
Zuckerberg's change in direction is seen as clashing with LeCun's vision. At FAIR, LeCun has focused on developing the next-generation "world model." A "world model" is an AI system designed to understand the physical world by learning from video and spatial data.
LeCun's position is that AI based on LLMs—which are essentially just next-word prediction technology—may be useful in daily life but cannot acquire the ability to reason and plan like humans. He presents a "world model," in which AI learns about the world by seeing and hearing directly and can further perform prediction and reasoning, as an alternative. LeCun projected that it will take about 10 years to complete this model.
LeCun is reported to be leaving Meta to establish an AI Start - Up, which is also expected to focus on world model research.
LeCun is a leading scholar known as one of the "Godfathers of AI," along with Geoffrey Hinton of the University of Toronto, who won the Nobel Prize in physics last year, and Yoshua Bengio of the University of Montreal. The three were recognized for their achievements in AI and won the Turing Award, known as the Nobel Prize of computing, in 2018.