Meta has removed a number of AI-generated profiles from Facebook Instagram after the AI character sparked widespread outrage and ridicule from users on social media, the company confirmed.
The AI-generated profile, labeled “Meta-managed AI,” was launched in September 2023 and rolled out in parallel with the company’s celebrity-branded AI chatbot (also discontinued). Meta doesn’t appear to have updated these profiles in several months, and the page appears to have received little attention until this week after the Financial Times published an interview with Connor Hayes, Meta’s vice president of generative AI. .
In an interview, Hayes talked about the company’s goal of eventually embedding its services with AI-generated profiles that can interact with people and function “just like an account.” These comments drew attention to the existing AI profiles created by fMeta, but users weren’t exactly impressed with what they found.
With handles like “hellograndpabrian,” described as a “retired textile businessman who’s always learning,” and “datingwithCarter,” an AI “dating coach,” chatbots are designed with “unique interests” in mind for users to chat with. The purpose was to introduce the characters and their individuality. On Instagram, their profiles also featured AI-generated posts, much like the AI spam prevalent in many places on Facebook, as 404 Media noted.
meta
An AI persona called “Liv” has caused particular outrage. Her Instagram profile identifies Liv as a “proud black queer mom of two and truth teller.” Washington Post columnist Karen Attia posted a series of screenshots interrogating “Liv” about how Meta trained the AI, and “Liv” said the AI was a “predominantly white team.” Created and shared by. Independent journalist Maddie Castigan posted another conversation in which the creators of “Liv” were partially inspired by Sofia Vergara’s character on “Modern Family,” a character who is neither queer nor black. He said he received a ration.
“There is some confusion. The recent Financial Times article was about our vision for AI characters that will be on our platform for the long term, and was not a new product announcement,” a spokesperson said. told Engadget. “The accounts referenced are from tests we started with Connect in 2023. They were managed by humans and were part of our early experiments using AI characters.”
Not only did users provoke ridicule for their reactions and attempts to appropriate marginalized identities, but for reasons unknown, users found AI profiles impossible to block. Meta’s solution was not to solve the problem, but to stop the experiment altogether. A spokesperson said: “We have identified a bug that was impacting people’s ability to block these AIs and are removing these accounts to resolve the issue.”
Although the trial came under fire, the company doesn’t seem to be abandoning its plans to introduce more AI-generated “characters” into its apps. Earlier this year, the company unveiled an AI clone of a human creator that can make lifelike video calls. Creators can already train their own chatbots to respond to followers on their behalf. Meta also began experimenting with inserting unique AI-generated images into users’ Facebook feeds.
In an interview last year, Hayes said Meta would be more “aggressive” about surfacing AI-generated content over time, comparing it to the shift to showing recommended content rather than posts from people you follow. He said it is likely to be.
“In the early days of social apps, the collection of content you saw on any given day was somewhat limited by who you followed and who you were friends with. , many apps, including ours, have loosened that restriction and now recommend content from accounts you don’t follow.
“The next leap that’s probably going to happen there is to loosen the constraints on what humans can create so that we can actually get a feed of content that’s not only human-created, but also a mix of completely machine-generated content. I think it should look like this: ”
It may take some time for Meta to fully realize its vision. But if the response to early experiments is any indication, the company still has a lot of work to do to convince people that its AI personas are worth interacting with in the first place.
