In blow to Google, AI research manager leaves after two colleagues ousted

Samy Bengio oversaw Timnit Gebru, a well-known artificial intelligence researcher fired by the tech company

FILE PHOTO: The Google Inc. logo is seen outside their headquarters in Mountain View, California August 18, 2004. REUTERS/Clay McLachlan/File Photo

Google research manager Samy Bengio, who oversaw the company’s AI ethics group until a controversy resulted in two female leaders being ousted, resigned on Tuesday to pursue other opportunities.

Mr Bengio, who managed hundreds of researchers in the Google Brain team, announced his departure in an email to staff that was obtained by Bloomberg. His last day will be April 28.

An expert in a type of AI known as machine learning, Mr Bengio joined Google in 2007.

Ousted Ethical AI co-leads Timnit Gebru and Margaret Mitchell had reported to Mr Bengio and considered him an ally.

In February, Google reorganised the research unit, placing the remaining Ethical AI group members under Marian Croak, cutting Mr Bengio’s responsibilities.

“While I am looking forward to my next challenge, there is no doubt that leaving this wonderful team is really difficult,” Mr Bengio wrote in the email.

He did not refer to Ms Gebru, Ms Mitchell or the disagreements that led to their departures. Google declined to comment.

In November, Mr Bengio’s then-manager Megan Kacholia met with Ms Gebru to demand she retract a paper co-written with Ms Mitchell and other Google researchers that criticised an AI technology that powers some of Google’s search results.

Google fired Ms Gebru and Ms Mitchell in December and February, respectively.

“The resignation of Samy Bengio is a big loss for Google,” said El Mahdi El Mhamdi, a scientist at Google Brain.

He said Mr Bengio helped build “one of the most fundamental research groups in industry since Bell Labs, also one of the most profitable ones.”

In his email, Mr Bengio said he “learnt so much with all of you, in terms of machine-learning research of course, but also on how difficult yet important it is to organise a large team of researchers so as to promote long-term ambitious research, exploration, rigour, diversity and inclusion”.

Before joining Google, Mr Bengio helped to develop Torch, an open-source framework and package of tools for the development of machine-learning algorithms.

At Google, he was part of the TensorFlow team, building a rival offering that surpassed Torch in popularity.

Facebook researchers used his earlier work for its PyTorch library of AI tools.

Mr Bengio also published research in areas like adversarial machine learning, which feeds false or misleading information to algorithms to try to trick or corrupt them.