Google Fires Another AI Researcher for Questioning Findings, Company Says Otherwise


Google fired an AI researcher who called into question a well-known paper the company published in 2020, a third high-profile termination in less than two years, although the company says his removal was for other reasons.

Satrajit Chatterjee was fired in March after he and his team challenged a Google research paper about the use of artificial intelligence to make computer chips, according to The New York Times. The research paper, which was published in Nature, a prestigious science journal, said computers could design parts of chips faster than humans. 

In an internal email, Chatterjee, 43, challenged some of the paper’s assertions, according to the Times. He also questioned whether the technology had been fully tested, the paper reported.

In a statement to the Times, Google defended its research paper and practices.  

“We thoroughly vetted the original Nature paper and stand by the peer-reviewed results,” said Zoubin Ghahramani, a vice president at Google Research in the statement to the Times. “We also rigorously investigated the technical claims of a subsequent submission, and it did not meet our standards for publication.”

A tweet by Times reporter Daisuke Wakayabashi made mention of one of the paper’s author’s being harassed by Chatterjee. 

Google did not divulge any additional details or context regarding the nature of this harassment, but said he was fired for cause. 

“We’re committed to creating an inclusive research environment, so we firmly uphold our standard for respectful discourse among our researchers,” a Google representative said in a statement. 

Chatterjee’s dismissal follows two high-profile firings within Google AI. In December 2020, the Google unit fired Timnit Gebru, one of the leaders of its Ethical AI team after she published a paper identifying the dangers of bias in AI language models. Two months later, Margaret Mitchell, who co-led the Ethical AI team, was fired for violating the company’s code of conduct for reportedly using automated software to scan her messages to find discriminatory treatment of Gebru.

Google has said Gebru published the paper without going through the proper channels. 

The search giant has made its AI research central to the company’s reputation for innovation. Alphabet CEO Sundar Pichai has said AI will be more important than electricity or fire. The company has integrated AI models in its core search product as a way to give users better results while lowering spam

Google rejected a paper Chatterjee and his team wrote that rebuts the original paper, according to the Times. The researchers attempted to escalate the issue to Pichai and Alphabet’s board, the paper reported. 

“In line with our AI Principles, we also rigorously investigated the technical claims of a subsequent submission, and it did not meet our standards for publication,” said Ghahramani. He went on to say that Google Research is growing and plans to hire more this year than last.


Source link