A strange experiment. Meta wanted to fill Facebook and Instagram with AI clones, but changed its mind

The company Meta has destroyed a bunch of people's own profiles on Facebook and Instagram. All of these pages were completely generated by artificial intelligence.

The AI-generated profiles, which were labeled «AI-powered by Meta», appeared in September 2023. Meta hadn’t updated any of them for months, and they went unnoticed until Meta’s VP of Generative AI, Connor Haze, mentioned them in an interview with the Financial Times. Haze spoke about them while discussing the company’s goal of eventually populating its services with AI-generated profiles that can interact with people and function just like regular accounts. Examples of such pages include hellograndpabrian (a retired businessman), datingwithCarter (a dating coach), himamaliv (a black queer person with two children), and more. On Instagram, their profiles also featured posts generated by AI. However, these pages were run by real people.

«The accounts linked to are from a test we ran on Connect in 2023. They were run by people and were part of an early experiment we were running with AI characters», — a Meta spokesperson said in a comment to Engadget.

Although these AI profiles were not a new product for the company, the mention of them caused a wave of outrage and mockery on social media. In addition, users found it impossible to block these AI-generated pages. This is what the company used as an excuse to remove the experimental pages.

«We discovered a bug that was affecting people ability to block these AIs, and we are removing these accounts to resolve the issue», — said a spokesperson.

Despite the failure of this experiment, the company doesn't seem to be giving up on its plans to add more AI-created characters. Earlier this year, the company introduced AI clones for celebrities that can make realistic video calls and communicate with followers on their behalf.