Book App Alarmed as Its AI Starts Mocking Users for Reading Books by Women and Minorities

Yikes.

Page Burner

The popular book app Fable has come under fire for its AI-generated annual roundups, which some users say gave them offensive messages about race and gender, Wired reports.

These days, every app imaginable is deploying its own take on the wildly popular Spotify Wrapped feature by providing personalized end-of-year recaps for users' consumption habits. Hopping on the trend, Fable tried to stand out by using AI to "playfully roast" its readers.

In some cases, however, the AI veered into overly edgy territory — in what appears to be an all-too-common case of a large language model defying its guardrails.

The summary for Fable user Tiana Trammell, for example, praised her for being a "soulful explorer" of "Black narratives," before about-facing completely: "Don't forget to surface for the occasional white author, okay?"

"I typically enjoy my reader summaries from [Fable], but this particular one is not sitting well with me at all," Trammell wrote on Threads.

Become the Woker

Other users reported similar cases of the AI striking an almost comically reactionary tone.

One writer's Fable summary called him a "diversity devotee" and asked if he's "ever in the mood for a straight, cis white man's perspective." Trammell says she's also seen other people whose summaries snidely commented on "disability and sexual orientation."

It's unclear how widespread these cases were. But Fable has responded to the complaints by issuing a formal apology. "To our community: we are deeply sorry for the hurt caused by some of our Reader Summaries this week," the company posted on Threads. "We promise to do better."

What "doing better" looks like, if you're wondering, is tinkering with its AI model. "For the time being, we have removed the part of the model that playfully roasts the reader, and instead, the model simply summarizes the user's taste in books," Kimberly Marsh Allee, Fable's head of community, told Wired.

Closed Book

Some users, though, would prefer to see the AI's proverbial head roll.

"They need to say they are doing away with the AI completely," fantasy romance author A.R. Kaufer told Wired. "This 'apology' on Threads comes across as insincere, mentioning the app is 'playful' as though it somehow excuses the racist/sexist/ableist quotes."

Fable's decision to muzzle its AI raises serious questions about the tech's usefulness for major brands. Surely AI's novelty is that it can whip up pithy remarks that sound like human speech. If it's not trustworthy enough to perform that function without making a serious faux pas at every turn, then why should brands trust it with their customers? And doesn't limiting these bots to become dry summarizers defeat the point of what makes them appealing in the first place?

More on AI: Meta Kills AI-Generated "People" Like "Proud Black Queer Momma"