Google employees who worked with Timnit Gebru are coming out publicly to dispute claims against the star AI ethics researcher. On Monday, the team published a letter on the Google Walkout Medium account firmly stating that Gebru was fired and did not resign as Google’s head of artificial intelligence, Jeff Dean, said. They also said the publication review policy that Gebru was supposed to follow was applied “unevenly and discriminatorily.”
“Dr. Gebru’s dismissal has been framed as a resignation, but in Dr. Gebru’s own words, she did not resign,” the letter says. It notes that Gebru asked for certain conditions to be met in order for her to stay at Google, including transparency around who wanted her paper retracted. Ultimately, the leaders of the ethical AI team said they could not meet these conditions and preemptively accepted her resignation. Her own manager said he was “stunned.”
The paper that got Gebru fired detailed potential risks associated with large language processing models, including over-relying on data from wealthy countries that have more internet access. “The result is that AI-generated language will be homogenized, reflecting the practices of the richest countries and communities,” wrote MIT Technology Review.
This research could have been problematic for Google, which created a large language model called BERT in 2018, which has changed how it queries search results.
Gebru was planning to present the paper at a computer science conference in March. On October 7th, she submitted it for review internally at Google. Shortly after midnight on October 8th, it was approved.
In his statement, Dean said the research team requires two weeks for review. “Unfortunately, this particular paper was only shared with a day’s notice before its deadline,” he wrote.
But Gebru’s team is pushing back on that assessment, saying the review policy is meant to be flexible, and most people do not follow the structure Dean laid out. The team collected data showing the vast majority of approvals happen right before the deadline, and 41 percent happen after the deadline. “There is no hard requirement for papers to actually go through this review with two weeks notice,” they wrote.
Google managers asked Gebru to retract the paper or take her name off it, a request she said felt like censorship in an interview with Wired. “You’re not going to have papers that make the company happy all the time and don’t point out problems,” she said. “That’s antithetical to what it means to be that kind of researcher.”
Gebru sent an email to the Brain Women and Allies listserv at Google detailing the pushback she’d gotten on the paper. She also voiced exasperation with the company’s diversity, equity, and inclusion efforts. “The DEI [objectives and key results] that we don’t know where they come from (and are never met anyways), the random discussions, the ‘we need more mentorship’ rather than ‘we need to stop the toxic environments that hinder us from progressing’ the constant fighting and education at your cost, they don’t matter,” she wrote.
This frustration was shared by members of her team who felt the company’s goals to create a more diverse and equitable workplace were weak. “They’re really paltry demands,” says Alex Hanna, a senior researcher who worked under Gebru.
More than 1,500 employees have signed a Google Walkout petition protesting Gebru’s dismissal. “Instead of being embraced by Google as an exceptionally talented and prolific contributor, Dr. Gebru has faced defensiveness, racism, gaslighting, research censorship, and now a retaliatory firing,” the petition says.
But it’s members of her team who feel the loss most acutely. “We’re pretty deflated,” Hanna says. “When someone who is the heart and soul of your team gets fired ostensibly for doing ethics research, what can you do?”
Correction: An earlier version of this article stated the letter was signed by more than 1,500 Google employees. That was the Google Walkout petition. We regret the error.