Google AI loses another leader following firing of ethics researchers

7 Apr 2021

Image: © JHVEPhoto/Stock.adobe.com

Samy Bengio, a leader from Google’s AI research team, will leave the company later this month.

Samy Bengio has left Google Brain, Google’s deep learning AI research team, Bloomberg reports.

A research scientist at Google since 2007, Bengio will continue to lead a group of researchers on the Google Brain team until his departure on 28 April. According to Bloomberg, a reorganisation of this team announced in February had cut some of Bengio’s responsibilities.

Google vice-president Marian Croak was named head of responsible AI research at the company, tasked with making sure Google’s work in AI is responsible and has a positive impact. Croak took leadership of a new centre of expertise on responsible AI within Google Research, managing 10 teams and reporting directly to Jeff Dean, the head of Google AI and one of the originators of Google Brain along with Andrew Ng.

Dean has been the subject of much criticism following the abrupt dismissal of Dr Timnit Gebru as co-lead of Google’s ethical AI team late last year. Shortly after the appointment of Croak, Google fired AI ethics lead Margaret Mitchell, who had worked alongside Gebru and was reportedly put under investigation for using a script to sift through company emails looking for examples of discrimination against Gebru.

Bengio did not cite the firing of Gebru or Mitchell in a farewell letter to staff seen by Bloomberg. However, his departure has been linked to the ongoing controversy at Google AI and Bengio previously posted a statement on Facebook in support of Gebru. Following her firing in December 2020, Bengio wrote: “I have always been and will remain a strong supporter of her scientific work.”

He is now leaving Google to pursue other opportunities, according to Bloomberg.

Unrest in Google AI

The discord in the ethical AI team is reported to have begun with a paper co-written by Gebru, which raised many issues around training language models on the data gleaned from the broadest segments of the internet.

“Researchers have sought to collect all the data they can from the internet, so there’s a risk that racist, sexist and otherwise abusive language ends up in the training data,” wrote Karen Hao, senior AI reporter at MIT Technology Review, who has seen the paper co-authored by Gebru.

According to The New York Times, Google demanded that Gebru either retract her name from the paper or pull it entirely, which she refused. Dean wrote that Gebru’s paper “ignored too much relevant research” that he claimed has shown improvements in AI made in recent years.

Prior to Croak’s appointment, David Baker, an engineering director at the company, and software developer Vinesh Kannan both left Google, citing the treatment of Gebru as the reason behind their departure.

Baker, who had been at the company for 16 years, said that “we cannot say we believe in diversity, and then ignore the conspicuous absence of many voices from within our walls”.

Kannan also cited the “mistreatment” of recruiter April Christina Curley in a tweet about leaving the company. Curley, whose employment was also terminated by Google, has detailed the difficulties she experienced as a diversity recruiter at the company.

Since the February restructure, unrest at Google AI has continued. In March, CNN Business reported that Luke Stark, an assistant professor in AI ethics at Canada’s Western University, turned down a $60,000 Google Research Scholar award “to show his support for Gebru and Mitchell, as well as those who yet remain on the ethical AI team at Google”.

CNN also reported that AI conference organisers were rethinking Google’s sponsorship of their events.

Prior to that, University of Texas computer science researcher Vijay Chidambaram tweeted that he would no longer apply for Google funding until the company “change their ways”. This was in response to a tweet from Prof Hadas Kress-Gazit, a robotics researcher at Cornell University, who said she had emailed Google to cancel her participation in a machine learning and robot safety workshop.

“The reason is that I cannot support anything that has to do with research and Google after the whole [Timnit Gebru, Margaret Mitchell] and responsible AI team fiasco,” she wrote.

Elaine Burke is the host of For Tech’s Sake, a co-production from Silicon Republic and The HeadStuff Podcast Network. She was previously the editor of Silicon Republic.

editorial@siliconrepublic.com