Meta’s Top AI Scientist Is Reportedly Quitting to Build His Own Startup

Meta’s AI team might have just taken another major hit. The tech giant’s chief artificial intelligence scientist, Yann LeCun, will be leaving Meta in the coming months to launch his own start-up, according to the Financial Times.

LeCun is a big deal in AI. As a Turing Award winner, the scientist is considered one of the leading figures in modern AI. The Financial Times reported on Tuesday that LeCun is in early funding talks for a new venture.

The news, if true, is only the latest in a series of blows the tech giant has received in the past few months as it struggles to pan out its ambitious AI goals.

Meta CEO Mark Zuckerberg envisioned a big AI turnaround story after admitting that the company had fallen behind peers in the AI race. That intended turnaround started earlier this year with the formation of Meta Superintelligence Labs, for which the tech giant spent billions of dollars to poach top talent from OpenAI, Apple, and more.

The move also included pseudo-acquiring Scale AI by gutting talent at the startup and bringing over founder Alexandr Wang to lead Meta’s superintelligence team. According to previous reports, the 28-year-old tech executive’s leadership style has clashed with some employees. LeCun used to report to Meta’s chief product officer Chris Cox until the acqui-hire, but is now reporting to Wang, the FT reported on Tuesday.

Despite the flashy spending goals, things took a turn in August. In a surprise decision, Meta split its superintelligence division into four smaller groups only two months after Zuckerberg announced its formation. A few weeks after that, reports came out that Meta was already bleeding top AI talent in its superintelligence team. According to the report, at least three AI researchers had resigned after less than a month of employment at Meta. Then last month, the company went through yet another reorganization by cutting roughly 600 positions from its AI team.

While that was happening, Meta’s AI efforts keep stalling. At best, the company’s AI products either had their release dates delayed or fared worse than expected with users. At worst, the products have been riddled with controversy.

Meta made headlines in June after it was revealed that user prompts on the Meta AI app were publicly visible to others. Later in the summer, the company came under fire and found itself in the middle of a Senate probe after a Reuters report found that Meta allowed its AI chatbots to engage in “sensual” conversations with minors. Texas attorney-general Ken Paxton’s office has also opened an investigation of its own into Meta’s chatbots, this time over claims that it has impersonated licensed mental health professionals.

Meta’s AI chatbot “Big sis Billie” also caused public outrage in August when it invited a cognitively impaired New Jersey retiree to come meet “her” at a nonexistent New York apartment, and the man died on his way into the city.

For what it’s worth, Meta is dedicated to continuing to spend eye-watering figures in hopes of delivering on its ambitious AI promises (creating some form of superintelligence). But dedication and spending don’t always guarantee success. Meta’s last scheme, the D.O.A. Metaverse, is just one prime example of that.