Meta
AI research lab (formerly Facebook Artificial Intelligence Research, or FAIR) is “slowly dying,” some insiders say. Meta prefers to call it a “new beginning,” Fortune writes.
Image source: Steve Johnson/unsplash.com
The news that Meta’s vice president of AI research, Joelle Pineau, who led Meta AI for the past two years, has left many people surprised. FAIR was once at the forefront of AI development. But after Mark Zuckerberg refocused the company over the past two years on generative AI products, the lab has been gradually sidelined in favor of more commercially focused AI teams, Fortune notes.
For example, the latest AI model, Llama, was created by a separate team at GenAI Meta, not FAIR. FAIR has been losing ground, with talented researchers leaving for competing companies and startups: more than half of the 14 authors of the original Llama research paper, published in February 2023, left the company within six months, and at least eight of its top researchers have left in the last year.
Former FAIR employees told Fortune that the lab was “dying a slow death.” From that perspective, Pino’s departure could be interpreted as a “death rattle.”
While the search for a new leader continues, Yann LeCun, Meta’s chief AI scientist and FAIR’s founder, has been appointed as the lab’s interim head. He denies that the lab is close to closing, telling Fortune, “This is definitely not the death of FAIR.” He says the lab is about to experience a new dawn, and the existence of a group working on generative AI products actually allows FAIR to refocus on long-term research in the field. “This is more of a new beginning, where FAIR will focus on an ambitious, long-term goal, which is what we call AMI (advanced machine intelligence),” LeCun said.
Pino also told Fortune that she remains “very enthusiastic about Meta’s overall work and strategy in AI.” She said there is still strong support for FAIR within the company, and that her departure was because she wanted to “refocus her energy before moving on to something new.”
However, seven former Meta employees interviewed by Fortune last week say the lab has been slowly scaling back in recent years and has been given less access to computing power than teams focused on generative AI.
When OpenAI’s ChatGPT chatbot was released in late November 2022, Meta was perceived as being seriously behind OpenAI, Anthropic, and Google in the generative AI space. However, FAIR helped Meta regain its footing in the AI market by developing Llama, an open-source, freely available generative AI model that could compete with solutions from other leading companies. FAIR then released Llama 2 in July 2023.
But by September of that year, the lab was already in trouble. As The Information reported at the time, many of the original FAIR researchers working on Llama had left the company due to a competition for computing resources with another research group, Meta, which was developing a competing model that was later abandoned. In January 2024, FAIR was restructured again, with Llama 4 now being released by Meta’s generative AI division, led by VP of generative AI Ahmad Al-Dahle. After that, a team within a newly created division led by Meta VP Manohar Paluri took over the work on Llama models, leaving FAIR on the sidelines.
According to former FAIR employees, the company is now shifting resources to programs focused on generative AI, whereas FAIR has traditionally been involved in a range of projects spanning a wide range of AI subfields, of which generative AI was just one.