Google revamps AI teams in wake of researcher’s departure

Google restructured its responsible artificial intelligence efforts to centralize teams under a single executive, Marian Croak, a move by the internet giant to stabilize groups working on ethics research and products after months of chaos.

Croak, a vice president of engineering, will be the Lead for the Responsible AI Research and Engineering Center of Expertise, she said in a YouTube video announcing her appointment Thursday. The Alphabet Inc. unit has sought to defuse employee rancor stemming from the acrimonious departure of a prominent Black researcher, Timnit Gebru. Croak, a Black Google executive who is currently focused on site-reliability matters, will report to Jeff Dean, the senior vice president of Google AI.

Croak acknowledged in the video that there’s a lot of “dissension” in the ethical AI realm right now, with researchers disagreeing on principles.

“Whose definition of fairness, or safety, are we going to use?” she said. “There’s quite a lot of conflict right now within the field, and it can be polarizing at times. And what I’d like to do is have people have the conversation in a more diplomatic way, perhaps, than we’re having it now, so we can truly advance this field.”

Croak will oversee the Ethical AI team that’s become the focus of intense scrutiny as well as employees on other fairness teams. These include people working on machine learning, computer-vision systems, natural language processing and those who engineer fairness products, a person familiar with the situation told Bloomberg News, which was first to report the changes. Megan Kacholia, who attracted employee criticism after dismissing Gebru, will no longer oversee these researchers, the person said.

Gebru responded to the news by writing on Twitter that Google was trying to “neutralize one Black woman with another.”

The crisis began in early December when Gebru, who’s best known for showing how facial recognition algorithms are better at identifying White people than Black people, said she was fired by email. Google said it had accepted her resignation after a conflict over an AI research paper critical of its technology that Google executives demanded Gebru retract or remove Google authors from. Her dismissal upset the Ethical AI research team she co-led, with members of her group taking to Twitter to publicly support her and criticize Google.

Two weeks later, a group of Google artificial intelligence researchers sent a sweeping list of demands to management calling for new policies and leadership changes. Five weeks ago, Google also sidelined the other leader of its AI ethics research team, Margaret Mitchell, locking her out of its corporate network.

This isn’t the first time Google has turned to Croak to handle the issue. A few days after Gebru’s dismissal, Croak moderated a meeting of Dean and Kacholia on one side, and researchers and the Black Googlers Network on the other.

“My biggest disappointment is with Marian Croak legitimizing this,” Gebru wrote on Twitter. “Starting with hosting Megan & Jeff’s gaslighting session.”