Feb 19, 2021 - Technology

Google tweaks diversity, research policies following inquiry

Google AI head Jeff Dean
Jeff Dean. Photo: Google

Google told employees Friday it has wrapped up its investigation into the ouster of prominent AI researcher Timnit Gebru. The company declined to say what the internal inquiry found, but said it is making some changes to how it handles issues around research, diversity and employee exits.

Why it matters: The treatment of Gebru, both before and after she was forced out of the company, has outraged people within Google's Ethical AI team and others inside and outside of the company. Google's handling of the matter has also raised questions about the company's commitment to diversity and to employing ethicists who are free to question the company's actions.

Under its new policies, Google says it will:

  • tie pay for those at the vice president level and above partly to reaching diversity and inclusion goals.
  • streamline its process for publishing research.
  • increase its staff related to employee retention.
  • enact new procedures around potentially sensitive employee exits.

"I understand we could have and should have handled this situation with more sensitivity," Google AI head Jeff Dean said in a memo on Friday, obtained by Axios, outlining the changes. "And for that, I am sorry."

I heard and acknowledge what Dr. Gebru’s exit signified to female technologists, to those in the Black community and other underrepresented groups who are pursuing careers in tech, and to many who care deeply about Google's responsible use of AI.  It led some to question their place here, which I regret.
— Google AI head Jeff Dean, in an internal email on Friday

Gebru is a top AI ethics researcher and one of a handful of prominent black women in the field, and Google had frequently promoted her work as a way to highlight its commitment to the AI ethics field. But the relationship had soured as her research became critical of the company's work, ultimately leading to her forced departure.

Between the lines: None of these moves seem likely to quell the continued frustration over this matter, especially within Google's Ethical AI team, members of whom says they continue to be blindsided by the company's moves.

  • Earlier this week Google named Marian Croak to head ethical AI work throughout Google's research organization. Croak is a longtime Google employee, a Black woman and a prominent researcher, having done pioneering work in the field of handling voice over the Internet; however, none of her work relates directly to ethical AI.
  • Her hiring was first reported Wednesday night by Bloomberg News, ahead of an announcement to staff the next day.

Meanwhile, one of Gebru's colleagues, Margaret Mitchell, remains locked out of her corporate e-mail. Google said last month it was investigating why Mitchell downloaded a large number of files and shared them with people outside the company.

  • According to a source, Mitchell had been using automated scripts to look through her messages to find examples showing discriminatory treatment of Gebru before her account was locked.
  • The company declined on Friday to offer an update on Mitchell's status, but as of now she remains a company employee.

"Research leadership has talked a big game about accountability," ethical AI researcher Alex Hanna said in a tweet on Thursday. "But there is zero accountability for anything that's happened for the past three months. Research leadership doesn't know the meaning of the word, as far as I'm concerned."

Go deeper