Facebook AI Research (FAIR): History, Teams, Projects, and Vision

Facebook has transformed from a pure social‑media company into a major technology research center, investing in drones, virtual reality, and especially artificial intelligence through its Facebook AI Research lab (FAIR).

FAIR was launched in 2013 when Mark Zuckerberg, CTO Mike Schroepfer, and other executives decided to build a world‑class AI research organization, recruiting pioneers like Yann LeCun and Marc Aurelio Ranzato.

The initial team was small—about 30 researchers and 15 engineers—spread across New York, Menlo Park, and later Paris, collaborating with other Facebook groups such as language technology.

FAIR’s long‑term ambition is to achieve “general intelligence,” moving beyond narrow supervised learning toward unsupervised learning, reinforcement learning, and memory‑augmented neural networks that can reason about the world.

Key research directions include memory networks that combine convolutional neural networks with a differentiable memory, enabling the system to answer story‑based questions, and the “embedding world” project that learns vector representations of images, posts, comments, and videos to capture their relationships.

FAIR also produced practical breakthroughs like DeepFace, a facial‑recognition system reaching 97% accuracy, and a language‑technology group that provides translation for hundreds of language pairs, improving cross‑lingual communication for billions of users.

The Facebook M personal assistant, built on the acquisition of Wit.ai, demonstrates how AI can handle complex tasks beyond simple voice commands, with a human‑in‑the‑loop training process that gradually reduces the need for manual intervention.

FAIR’s research is openly published and often open‑sourced; many of its contributions, such as the Torch library, are shared with the broader AI community, emphasizing collaboration as essential to solving the grand challenge of general AI.