Less than two years after Google fired two researchers who criticized the biases built into artificial intelligence systems, the company fired a researcher who questioned an article it published about the capabilities of a specialized type of artificial intelligence that is being developed. used in making computer chips.
The researcher, Satrajit Chatterjee, led a team of scientists in challenging the famous research paper, published last year in the scientific journal Nature, and said computers could design certain parts of a computer chip faster and better than humans.
dr. Chatterjee, 43, was fired in March shortly after Google told its team it would not publish an article refuting some of the claims in Nature, said four people familiar with the situation who were not allowed to speak openly about the torment. to do. Google confirmed in a written statement that Dr. Chatterjee “was terminated for a reason”.
Google declined to comment on Dr. Chatterjee, but it offered a full defense of the research he criticized and of the reluctance to publish his review.
“We have thoroughly vetted the original Nature paper and stand behind the peer-reviewed results,” Zoubin Ghahramani, vice president at Google Research, said in a written statement. “We also thoroughly investigated the technical claims of a subsequent submission and it did not meet our publishing standards.”
The resignation of Dr. Chatterjee was the latest example of dissension in and around Google Brain, an AI research group believed to be key to the company’s future. After spending billions of dollars hiring top researchers and creating new kinds of computer automation, Google has grappled with a wide variety of complaints about how it builds, uses, and portrays these technologies.
The tension between Google’s AI researchers reflects much bigger problems in the tech industry, which faces countless questions about new AI technologies and the thorny social problems that have entangled those technologies and the people who build them.
Read more about artificial intelligence
The recent dispute also follows a well-known pattern of layoffs and dueling claims of misconduct among Google’s AI researchers, a growing concern for a company that has staked its future on infusing artificial intelligence into everything it does. Sundar Pichai, the CEO of Google’s parent company Alphabet, has likened AI to the advent of electricity or fire, calling it one of humanity’s most important endeavors.
Google Brain started more than a decade ago as a side project when a group of researchers built a system that learned to recognize cats in YouTube videos. Google executives were so taken by the prospect of machines being able to learn skills on their own that they quickly expanded the lab and laid a foundation to recreate the company with this new artificial intelligence. The research group became a symbol of the company’s greatest ambitions.
Before she was fired, Dr. Get permission to publish a research paper on how AI-based language systems, including technology built by Google, may end up using the biased and hateful language they learn from text in books and on websites. dr. Gebru said she had become annoyed with Google’s response to such complaints, including the refusal to publish the paper.
A few months later, the company fired the other head of the team, Margaret Mitchell, who publicly criticized Google’s handling of the situation with Dr. Gebru denounced. The company said that Dr. Mitchell had broken his code of conduct.
The Nature article, published last June, promoted a technology called reinforcement learning, which the paper said could improve the design of computer chips. The technology has been hailed as a breakthrough for artificial intelligence and a vast improvement on existing approaches to chip design. Google said it used this technique to develop its own chips for artificial intelligence computing.
Google had been applying the machine learning technique to chip design for years and published a similar article a year earlier. Around that time, Google asked Dr. Chatterjee, who has a doctorate in computer science from the University of California, Berkeley, and had worked as a research scientist at Intel, to see if the approach could be sold or licensed to a chip design company, people familiar with the case said. .
But dr. Chatterjee expressed reservations about some of the paper’s claims in an internal email, and questioned whether the technology had been thoroughly tested, three of the people said.
As the debate over that research continued, Google threw another paper at Nature. For the submission, Google made some edits to the earlier article and removed the names of two authors, who had worked closely with Dr. Chatterjee and also expressed concern about the article’s key claims, the people said.
When the newer article was published, some Google researchers were surprised. They believed it had failed to follow a publication approval process that Jeff Dean, the company’s senior vice president who oversees most of its AI efforts, said was necessary in the wake of Dr. Gebru, the people said.
Google and one of the paper’s two lead authors, Anna Goldie, who co-wrote it with a fellow computer scientist, Azalia Mirhoseini, said the changes from the earlier paper didn’t require the full approval process. Google allowed Dr. Chatterjee and a handful of internal and external researchers to work on a paper that questioned some of his claims.
The team submitted the rebuttal to a so-called resolution committee for approval for publication. Months later, the paper was rejected.
The investigators working on the rebuttal said they wanted to escalate the matter to Mr Pichai’s board of directors and Alphabet. They argued that Google’s decision not to publish the rebuttal violated its own AI principles, which include upholding high standards of scientific excellence. Shortly afterwards, Dr. Chatterjee was told he was no longer an employee, the people said.
Mrs. Goldie said that Dr. Chatterjee had asked to manage their project in 2019 and they declined. When he later criticized it, she said, he was unable to substantiate his complaints and ignored the evidence they presented in response.
“Sat Chatterjee has been campaigning misinformation against me and Azalia for over two years,” Ms Goldie said in a written statement.
She said the work was peer-reviewed by Nature, one of the most prestigious scientific publications. And she added that Google had used their methods to build new chips and that these chips were currently being used in Google’s computer data centers.
Laurie M. Burgess, the attorney for Dr. Chatterjee, said it was disappointing that “certain authors of the Nature article are trying to stop the scientific discussion by defaming and attacking Dr. Chatterjee for simply seeking scientific transparency.” Ms. Burgess also questioned Dr. Dean, one of 20 co-authors of the Nature paper.
“Jeff Dean’s actions to suppress the release of all relevant experimental data, not just data that support his favored hypothesis, should be deeply troubling to both the scientific community and the wider community using Google services and products,” said Mrs. Burgess.
dr. Dean did not respond to a request for comment.
After the rebuttal was shared with academics and other experts outside of Google, controversy spread across the global community of researchers specializing in chip design.
Chipmaker Nvidia says it has used chip design methods similar to Google’s, but some experts aren’t sure what Google’s research means for the larger tech industry.
“If this works really well, that would be really great,” said Jens Lienig, a professor at the Technical University of Dresden in Germany, referring to the AI technology described in Google’s article. “But it’s not clear if it works.”
