A Black woman known as a “rare voice of public criticism” at Google has been fired by the tech company after sending an internal email calling out the company’s treatment of minority employees—and in particular, those who are Black and female.
Timnit Gebru, who co-led the Ethical Artificial Intelligence team at Google, is considered a pioneer in the field, with a particular focus on how facial recognition software is biased against people of color. As Bloomberg reports, Gebru says she was fired by the head of Google’s AI division on Wednesday for an email she sent to coworkers in which she said she felt “constantly dehumanized” at the tech company.
Gebru says she believes that her firing was intended to have a chilling effect on other employees, sending a message about the consequences of speaking up.
The email followed a struggle between Gebru and the company over a research paper Gebru had co-written with six other people, including four Google employees, which was critical of AI systems known as “large language models,” the Washington Post reports.
Here’s how Bloomberg summed up the analytical paper, which had been approved by Gebru’s manager before being sent to others at the company for approval:
The paper called out the dangers of using large language models to train algorithms that could, for example, write tweets, answer trivia and translate poetry, according to a copy of the document. The models are essentially trained by analyzing language from the internet, which doesn’t reflect large swaths of the global population not yet online, according to the paper. Gebru highlights the risk that the models will only reflect the worldview of people who have been privileged enough to be a part of the training data.
Google wanted Gebru to retract the paper, she said. And if she wouldn’t do that, they at least wanted the names of Google employees stricken from the research.
“It’s the most fundamental silencing,” Gebru said in an interview about Google’s demands. “You can’t even have your scientific voice.”
The struggle over the research paper topped off years of mounting tensions between Gebru and the company, which Gebru described in her email to researchers on the company’s AI team under the mailing list “Brain Woman and Allies.”
The letter was published on tech writer Casey Newton’s newsletter, Platformer, on Thursday.
“There is zero accountability” at Google, Gebru wrote. “There is no incentive to hire 39% women: your life gets worse when you start advocating for underrepresented people.
“We just had a Black research all hands with such an emotional show of exasperation,” she continued. “Do you know what happened since? Silencing in the most fundamental way possible.”
Gebru detailed several incidents between her and the company, including a potential lawsuit she and another colleague had considered filing against Google.
“Google lawyers were prepared to throw us under the bus,” Gebru said. But after she and a colleague hired “feminist lawyers,” she was given “some random ‘impact award’” by the tech giant.
“Pure gaslighting,” Gebru said, ultimately advising her recipients to “stop writing your documents because it doesn’t make a difference.”
“There is no way more documents or more conversations will achieve anything.”
After sending the email, Gebru said her managers told her the message reflected “behavior that is inconsistent with the expectations of a Google manager,” then told her they were accepting her resignation. Except, according to Gebru, she had not offered to resign. Gebru, who was on vacation this week, found that the company had already terminated her access to company email accounts, the Post reports.
After sharing the news of her firing on Twitter, many of her colleagues in and outside of Google responded with messages of support.
As Bloomberg notes, Gebru is considered one of the tech industry’s “leading voices” in the ethical use of AI. One of the few Black women in her field, she is responsible for the 2018 landmark study showing facial recognition software misidentified dark-skinned women up to 35 percent of the time. The same technology “worked with near precision on White men,” writes the Post.
She also was not afraid to clash publicly with her superiors, taking issue with Jeff Dean, Google’s head of AI, after he tweeted support for law enforcement in May, less than a week after George Floyd was killed while being arrested by Minneapolis police officers.
“You might think something is perfectly fine but you’re not the one dealing with this. This adds to the violence against us,” Gebru replied on Twitter. “Most people would not even say anything to you.”
Her outspokenness came with consequences. Speaking to Bloomberg News in August, Gebru said that, while Black Google employees are heralded as examples of the company’s commitment to diversity, when they speak out about their work conditions, they are given excuses, ignored, or will have their tone policed. In Gebru’s case, the former Google employee said her reputation for speaking out on diversity issues was used against her. When the company was considering having her manage another employee, supervisors said they were concerned about whether she could manage someone else if she was so unhappy, Bloomberg reported.
“People don’t know the extent of the issues that exist because you can’t talk about them,” Gebru told the publication at the time. “And the moment you do, you’re the problem.”