Yann LeCun, a pioneering scientist who led Meta’s artificial intelligence research efforts for more than a decade, is leaving the company to create his own start-up, he said on Wednesday.
Dr. LeCun’s departure follows a shake-up at Meta in its A.I. efforts. In June, the company invested $14.9 billion in the A.I. start-up ScaleAI and made that company’s chief executive, Alexandr Wang, its new chief A.I. officer. Weeks later, Mr. Wang named Shengjia Zhao, a researcher whom Meta hired from rival OpenAI, as its chief scientist, the same title as Dr. LeCun.
His exit ends an era for Meta’s A.I. research division, FAIR, which Dr. LeCun had served as a founding director. His new start-up will focus on advanced machine intelligence, a kind of A.I. he researched at Meta and as a professor at New York University. Meta will work with the new company, Dr. LeCun said.
“The impact of FAIR on the company, on the field of A.I., on the tech community, and on the wider world has been spectacular,” Dr. LeCun, 65, wrote in a social media post. Advanced machine intelligence “will have far-ranging applications in many sectors of the economy, some of which overlap with Meta’s commercial interests, but many of which do not.”
A Meta spokesman confirmed Dr. LeCun was departing the company and declined to provide details on the partnership.
Mark Zuckerberg, Meta’s chief executive, hired Dr. LeCun in 2013 as Facebook, as the company was then called, began investing in A.I. In 2018, Dr. LeCun won the Turing Award, the highest honor in computer science, for his work on neural networks, which are seen as foundational to the field of machine learning.
Dr. LeCun’s philosophy toward A.I. has diverged from his colleagues at Meta in recent years. Mr. Zuckerberg has made reaching “superintelligence,” which is A.I. that would exceed the abilities of the human brain, a goal at the company. Mr. Zuckerberg has poured money into developing large language models, which are the A.I. models that underpin chatbots, to reach superintelligence.
But Dr. LeCun has said that large language models will never be smart enough to be considered superintelligent. For an A.I. model to become superintelligent, it would need to be able to understand the physical world, Dr. LeCun has said, while large language models are only as knowledgeable as their training data.
Dr. LeCun said his new start-up’s focus on advanced machine intelligence will create a kind of A.I. that is broader in scope than large language models.
Meta’s A.I. efforts have been in flux for months as Mr. Zuckerberg has pushed his company to keep up with rivals including OpenAI and Google. Last month, Meta laid off hundreds of employees in its A.I. division, including some of the most senior FAIR directors.
Eli Tan covers the technology industry for The Times from San Francisco.
The post Yann LeCun, a Pioneering A.I. Scientist, Leaves Meta appeared first on New York Times.
