One afternoon in late November of last year, Timnit Gebru was sitting on the couch in her San Francisco Bay Area home, crying.
Gebru, a researcher at Google, had just clicked out of a last-minute video meeting with an executive named Megan Kacholia, who had issued a jarring command. Gebru was the coleader of a group at the company that studies the social and ethical ramifications of artificial intelligence, and Kacholia had ordered Gebru to retract her latest research paper—or else remove her name from its list of authors, along with those of several other members of her team.
The paper in question was, in Gebru’s mind, pretty unobjectionable. It surveyed the known pitfalls of so-called large language models, a type of AI software—most famously exemplified by a system called GPT-3—that was stoking excitement in the tech industry. Google’s own version of the technology was now helping to power the company’s search engine. Jeff Dean, Google’s revered head of research, had encouraged Gebru to think about the approach’s possible downsides. The paper had sailed through the company’s internal review process and had been submitted to a prominent conference. But Kacholia now said that a group of product leaders and others inside the company had deemed the work unacceptable, Gebru recalls. Kacholia was vague about their objections but gave Gebru a week to act. Her firm deadline was the day after Thanksgiving.
This article appears in the July/August 2021 issue. Subscribe to WIRED.
Photograph: Djeneba AduayomGebru’s distress turned to anger as that date drew closer and the situation turned weirder. Kacholia gave Gebru’s manager, Samy Bengio, a document listing the paper’s supposed flaws, but told him not to send it to Gebru, only to read it to her. On Thanksgiving Day, Gebru skipped some festivities with her family to hear Bengio’s recital. According to Gebru’s recollection and contemporaneous notes, the document didn’t offer specific edits but complained that the paper handled topics “casually” and painted too bleak a picture of the new technology. It also claimed that all of Google’s uses of large language models were “engineered to avoid” the pitfalls that the paper described.
Gebru spent Thanksgiving writing a six-page response, explaining her perspective on the paper and asking for guidance on how it might be revised instead of quashed. She titled her reply “Addressing Feedback from the Ether at Google,” because she still didn’t know who had set her Kafkaesque ordeal in motion, and sent it to Kacholia the next day.
On Saturday, Gebru set out on a preplanned cross-country road trip. She had reached New Mexico by Monday, when Kacholia emailed to ask for confirmation that the paper would either be withdrawn or cleansed of its Google affiliations. Gebru tweeted a cryptic reproach of “censorship and intimidation” against AI ethics researchers. Then, on Tuesday, she fired off two emails: one that sought to end the dispute, and another that escalated it beyond her wildest imaginings.
The first was addressed to Kacholia and offered her a deal: Gebru would remove herself from the paper if Google provided an account of who had reviewed the work and how, and established a more transparent review process for future research. If those conditions weren’t met, Gebru wrote, she would leave Google once she’d had time to make sure her team wouldn’t be too destabilized. The second email showed less corporate diplomacy. Addressed to a listserv for women who worked in Google Brain, the company’s most prominent AI lab and home to Gebru’s Ethical AI team, it accused the company of “silencing marginalized voices” and dismissed Google’s internal diversity programs as a waste of time.
Relaxing in an Airbnb in Austin, Texas, the following night, Gebru received a message with a 😮 from one of her direct reports: “You resigned??” In her personal inbox she then found an email from Kacholia, rejecting Gebru’s offer and casting her out of Google. “We cannot agree as you are requesting,” Kacholia wrote. “The end of your employment should happen faster than your email reflects.” Parts of Gebru’s email to the listserv, she went on, had shown “behavior inconsistent with the expectations of a Google manager.” Gebru tweeted that she had been fired. Google maintained—and still does—that she resigned.
Gebru’s tweet lit the fuse on a controversy that quickly inflamed Google. The company has been dogged in recent years by accusations from employees that it mistreats women and people of color, and from lawmakers that it wields unhealthy technological and economic power. Now Google had expelled a Black woman who was a prominent advocate for more diversity in tech, and who was seen as an important internal voice for greater restraint in the helter-skelter race to develop and deploy AI. One Google machine-learning researcher who had followed Gebru’s writing and work on diversity felt the news of her departure like a punch to the gut. “It was like, oh, maybe things aren’t going to change so easily,” says the employee, who asked to remain anonymous because they were not authorized to speak by Google management.
Dean sent out a message urging Googlers to ignore Gebru’s call to disengage from corporate diversity exercises; Gebru’s paper had been subpar, he said, and she and her collaborators had not followed the proper approval process. In turn, Gebru claimed in tweets and interviews that she’d been felled by a toxic cocktail of racism, sexism, and censorship. Sympathy for Gebru’s account grew as the disputed paper circulated like samizdat among AI researchers, many of whom found it neither controversial nor particularly remarkable. Thousands of Googlers and outside AI experts signed a public letter castigating the company.
But Google seemed to double down. Margaret Mitchell, the other coleader of the Ethical AI team and a prominent researcher in her own right, was among the hardest hit by Gebru’s ouster. The two had been a professional and emotional tag team, building up their group—which was one of several that worked on what Google called “responsible AI”—while parrying the sexist and racist tendencies they saw at large in the company’s culture. Confident that those same forces had played a role in Gebru’s downfall, Mitchell wrote an automated script to retrieve notes she’d kept in her corporate Gmail account that documented allegedly discriminatory incidents, according to sources inside Google. On January 20, Google said Mitchell had triggered an internal security system and had been suspended. On February 19, she was fired, with Google stating that it had found “multiple violations of our code of conduct, as well as of our security policies, which included exfiltration of confidential, business-sensitive documents.”
Google had now fully decapitated its own Ethical AI research group. The long, spectacular fallout from that Thanksgiving ultimatum to Gebru left countless bystanders wondering: Had one paper really precipitated all of these events?