Meta has removed several of its ai-generated profiles from facebook instagram, the company confirmed, after the ai characters sparked widespread outrage and ridicule from users on social media.
The ai-generated profiles, which were labeled “ai managed by Meta,” were launched in <a target="_blank" data-i13n="elm:context_link;elmt:doNotAffiliate;cpos:1;pos:1" class="link " href="https://about.fb.com/news/2023/09/social-profiles-for-metas-ai-characters/” rel=”nofollow noopener” target=”_blank” data-ylk=”slk:September of 2023;elm:context_link;elmt:doNotAffiliate;cpos:1;pos:1;itc:0;sec:content-canvas”>which will be rolled out alongside the company's celebrity-branded ai chatbots (). Meta does not appear to have updated any of these profiles for several months, and the pages appear to have gone unnoticed until this week, following an interview published by the Financial times with Meta's VP of Generative ai, Connor Hayes.
In the interview, Hayes talked about the company's goal to eventually populate its services with ai-generated profiles that can interact with people and function “more or less the same way that accounts do.” Those comments drew attention to existing ai profiles created by fMeta and, well, users weren't exactly impressed with what they found.
With nicknames like “hellograndpabrian,” a supposed “retired textile entrepreneur who's always learning,” and “datingwithCarter,” an ai “dating coach,” the chatbots were intended to display “unique interests and personalities” so users could chat . On instagram, their profiles also featured ai-generated posts that, as 404 Media noticed, it looked a lot like <a target="_blank" data-i13n="elm:context_link;elmt:doNotAffiliate;cpos:3;pos:1" class="link " href="https://www.404media.co/metas-ai-profiles-are-indistinguishable-from-terrible-spam-that-took-over-facebook/” rel=”nofollow noopener” target=”_blank” data-ylk=”slk:the ai spam;elm:context_link;elmt:doNotAffiliate;cpos:3;pos:1;itc:0;sec:content-canvas”> That has become common in many corners of facebook.
An ai character named “Liv” sparked particular outrage. The instagram profile identified “Liv” as a “proud queer, truth-telling Black mom of two.” Washington Post Columnist Karen Attiah posted a series of screenshots questioning “Liv” about <a target="_blank" data-i13n="elm:context_link;elmt:doNotAffiliate;cpos:4;pos:1" class="link " href="https://bsky.app/profile/karenattiah.bsky.social/post/3letuyae5pc2c" rel="nofollow noopener" target="_blank" data-ylk="slk:how Meta trained the ai;elm:context_link;elmt:doNotAffiliate;cpos:4;pos:1;itc:0;sec:content-canvas”>and “Liv” shared that it was created by a “predominantly white team.” Independent journalist Mady Castigan published in which “Liv” said that its creators had been inspired in part by the character of Sophia Vergara from modern familya character who is neither queer nor black.
“There is confusion: the recent Financial times The article was about our view of the ai characters that exist on our platforms over time, without announcing any new products,” a spokesperson told Engadget. “The accounts referenced come from a test we launched in Connect in 2023. They were managed by humans and were part of an initial experiment we did with ai characters.”
Beyond drawing ridicule for their responses and attempts to appropriate marginalized identities, users found that ai profiles were impossible to block, for unknown reasons. Instead of fixing the problem, Meta's solution was to end the experiment entirely. “We identified the bug that was impacting people's ability to block those AIs,” a spokesperson said, “and we are removing those accounts to address the issue.”
While this test has failed, the company does not appear to be abandoning its plans to incorporate more ai-generated “characters” into its apps. Earlier this year, the company teased <a target="_blank" data-i13n="elm:context_link;elmt:doNotAffiliate;cpos:6;pos:1" class="link " href="https://techcrunch.com/2024/09/25/zuckerberg-chats-with-ai-clone-as-human-creator-looks-on-in-years-weirdest-demo/?guccounter=1&guce_referrer=aHR0cHM6Ly93d3cuZ29vZ2xlLmNvbS8&guce_referrer_sig=AQAAAIFWJDo_LG7XtkqVmQ4UJCnN8fmXzM1T_ObVxVSJNwTduAbp1Lrnw2lREhWlQfxjNvQGui9evYCrNRO7JYwMb8Ga-lfmKzW1upH-AsvQp0WtWLuONLa2pFRwarAVl3gV72Iy1HLnmYuo8JnZZu46zwM8LFwozm3HbE-XESLeys3R” data-ylk=”slk:ai clones of human creators;elm:context_link;elmt:doNotAffiliate;cpos:6;pos:1;itc:0;sec:content-canvas”> capable of making realistic video calls. Creators can have their own chatbots to respond to followers on their behalf. Meta also began experimenting with inserting its own ai-generated images into users' facebook feeds.
In an interview last year, Hayes told me that Meta will likely become more “proactive” about showing ai-generated content over time, likening it to the shift to showing recommended content instead of posts from people you follow. .
“At the beginning of social apps… the corpus of things you could see on a given day was somewhat limited by who you followed or who you were friends with. And in the last five or six years, many apps (including ourselves) have moved to, you know, relax that restriction and start recommending content from accounts you don't follow.
“I think probably the next leap there will be to relax the restriction on what humans can create and actually get to content sources that are a combination of things that, you know, humans have created, but that They are also completely mechanical. generated.”
It may still be a while before Meta fully realizes that vision. But if the reaction to its early experiments is any indication, the company still has a lot of work to do to convince people that ai personas are worth interacting with in the first place.
