A leading Google researcher on artificial intelligence ethics said she was fired after criticizing the Alphabet Inc.-controlled company’s diversity initiatives. Google responded on Thursday to the recent friction between the internet giant and worker activists.
Timnit Gebru, who is black, said on Twitter that she was fired on Wednesday after sending an email to coworkers expressing frustration with the gender diversity guidelines in Google’s AI department. She also asked if the company’s bosses scrutinized their work more closely than people from other backgrounds. Gebru co-founded the nonprofit Black in AI [Negros em IA]who aims to increase the representation of black people in the artificial intelligence field and co-authored an important article on bias in facial analysis technology.
Jeff Dean, head of Google’s AI department, told employees in an email investigated by Reuters that Gebru had threatened to resign unless he knew which colleagues thought a draft she had written was unpublishable. Dean declined the request.
“We accept and respect his decision to step down from Google,” wrote Dean in the email, adding, “We all sincerely share Timnit’s passion for making AI more equitable and inclusive.”
Gebru said on a series of Twitter posts that Google cut it out of their systems without warning her or speaking to her about its concerns.
Gebru’s sudden departure leads to years of agony, including multiple resignations and layoffs in the AI department and other Google organizations, related to the company’s diversity and efforts to minimize the potential harm to its services.
More than 150 employees supported Gebru and asked Google to step up its commitment to academic freedom and explain why it decided to “censor” its article, according to a petition published online. Sherlyn Ifill, president of the National Association for the Advancement of People with Color (NAACP), President of the Legal and Educational Defense Fund, wrote on Twitter that Gebru’s resignation was “totally annoying” and “a disaster”.
Also on Wednesday (2), the National Council for Labor Relations published a complaint accusing Google of illegally monitoring and questioning several workers who were later dismissed for protesting company policies and attempting to organize a union.
Gebru’s article argued that technology companies could do more to ensure that AI systems designed to mimic writing and human language don’t reinforce historical gender bias and the use of offensive language, according to a draft by Reuters.
In his email sent to employees, Dean said the item was not delivered to the company in time for review and was presented at a conference without Google’s permission.
He also questioned some of his conclusions, which he believed to be based on outdated concerns, including the environmental impact of large numbers of computers processing data.
In response to the company’s rejection of his work, Gebru wrote on Twitter last week: “Nothing like a bunch of privileged white men trying to break the research done by marginalized communities into marginalized communities and command them to speak to ZERO. ” The level of disrespect is incredible “.
Google declined to comment on Gebru’s firing in addition to Dean’s email, which was first reported by technology news site Platformer.
Gebru previously worked at Microsoft Research and was the co-author of a highly cited 2018 article that found higher error rates in facial analysis technology for women with darker skin tones.
His new article, which was co-authored with people who don’t work at Google, is due to be presented at a computer science conference in March, according to a person familiar with the matter.
Translation by Luiz Roberto Mendes Gonçalves