In 2013, Facebook brought on board French AI pioneer Yann LeCun to establish its AI Research lab, a move that demonstrated the company's commitment to vying in the technologies reshaping the internet. The social network's prominent AI capabilities transformed an avalanche of photos, pokes, and personal data into a globally dominant platform.
However, backlash concerning election-interference advertisements, misinformation, and the platform's effects on mental well-being—issues Facebook hopes AI can address—has raised doubts about the company's ability to match its competitors in the cutthroat arena of AI research and development.
This week, Facebook reorganized its AI leadership, moving LeCun into a narrower position focused on AI strategy, guidance, and external advocacy. To oversee the technology's expansion, the company appointed Jérôme Pesenti, former head of IBM's Watson AI system, as its new vice president of AI.
This leadership shift underscores both the swift progress of AI within Facebook and the considerable distance the company still has to cover.
"Facebook has become a major force in AI, which it absolutely was not half a decade ago," remarked Pedro Domingos, a University of Washington professor, AI researcher, and author of "The Master Algorithm." "That said, compared to Google or Microsoft, they remain a minor player."
Domingos noted that Facebook's AI research team, numbering about 100, is just a tiny portion of the workforce at Google or Microsoft and has a considerably narrower focus. "This is the Red Queen hypothesis," he explained, invoking an evolutionary concept. "It's not about your absolute speed but, relatively speaking, how fast you move relative to others."
This week, Facebook announced plans to double the capacity of its Paris AI lab. Overall, the company now has over 100 AI researchers spread across the United States, Montreal, Tel Aviv, and Paris.
According to company spokesperson Ari Entin, LeCun's updated position mirrors the growing maturity of Facebook's research and product divisions.
"The truth is that AI has never been more crucial to Facebook," Entin stated. "Our teams are expanding. We are publishing and open-sourcing more than ever, and we're integrating AI across the platform at an exceptional scale."
LeCun directed inquiries to Facebook, and the company declined to make either LeCun or Pesenti available for interviews. In a Facebook post, LeCun wrote that his new responsibilities would center "on scientific leadership, strategy and external communication, with less emphasis on operational management."
Artificial intelligence has long served as the foundation for essential Facebook features that attract new users and maintain engagement, including facial recognition for photo tagging and the algorithms determining post placement on users' News Feeds.
Facebook CEO Mark Zuckerberg himself recruited LeCun, a New York University professor renowned for his deep learning innovations.
Facebook's initial AI lab was modest, yet Zuckerberg expressed lofty aspirations. "One of our goals for the next five to 10 years," he stated in 2015, "is to basically get better than human level at all of the primary human senses: vision, hearing, language, general cognition."
AI came to occupy an expanding share of Facebook's fundamental technologies, covering face recognition, language translation, ad targeting, video captioning, identifying objectionable content, and suggesting "people you may know."
"Yann has built a world-class team at Facebook, having it both publish great papers and ship products," remarked Andrew Ng, co-founder of Google Brain and former chief scientist at Chinese tech giant Baidu, citing what he described as notable achievements in natural language processing and image recognition.
However, AI is also at the heart of several of Facebook's most contentious and prominent challenges: handling misinformation and hate speech, preventing discriminatory advertising, and addressing concerns over attention manipulation and social media addiction.
Yet some specialists argue that Facebook's AI teams, similar to those at other major tech firms, have grappled with the inherent conflict between research groups—focused on long-term academic goals—and application teams tasked with developing functional features.
Hilary Mason, vice president of research at Cloudera and founder of machine-learning research company Fast Forward Labs, noted that very few organizations of Facebook's size have managed to strike an ideal equilibrium between research and practical outcomes.
She further remarked that Facebook continues to be highly attractive to AI professionals: "Data science types are attracted to where the data is, and Facebook has some of the most interesting data."
Facebook vies for talent with tech behemoths like Amazon and Google's DeepMind, which provide researchers opportunities to operate beyond a social media context and delve into voice assistants, healthcare, gaming, drone delivery, and other ambitious projects.
In 2016, LeCun mounted his own defense on the Q&A platform Quora, acknowledging that Google "is probably ahead" of Facebook and others in applying deep learning and other AI methods, but adding that "we are ambitious in our goals, we are here for the long run, and we have an impact on the company, which makes it easy for us to justify our existence."
Within Facebook, Pesenti will supervise both the division formerly led by LeCun—Facebook AI Research—and the Applied Machine Learning division, which creates and rolls out the AI that serves over 2 billion users globally each month.
Domingos emphasized that AI is vital for maintaining the platform's success, user engagement, and vitality.
"Facebook has this amazing business where they don't even have to troll the Web for content. People just upload their stuff and then they serve it back out with ads attached, and they print money. It's great to be Facebook," Domingos said. But its "machine learning has to respond. And if it doesn't respond, the whole site will be in much worse shape."
