Less than two years after dismissing two researchers who criticized the biases built into artificial intelligence systems, the company has fired a researcher who questioned a paper it published.
A team of scientists challenged a research paper that claimed computers were able to design parts of a computer chip faster and better than humans.
Four people familiar with the situation who were not permitted to speak openly on the matter said that Dr. Chatterjee was fired in March after his team was told that it would not publish a paper that rebutted some of the claims made in Nature. In a written statement, the company confirmed that Dr. Chatterjee had been terminated with cause.
The dismissal of Dr. Chatterjee was not explained by Google, but it offered a full-throated defense of the research he criticized and of its refusal to publish his assessment.
We thoroughly checked the original Nature paper and stood by the results.
The dismissal of Dr. Chatterjee was the latest example of disagreements within the research group. After spending billions of dollars to hire top researchers and create new kinds of computer automation, Google has struggled with a wide variety of complaints about how it builds, uses and portrays those technologies.
Tension among the A.I. researchers at Google is indicative of larger struggles in the tech industry, which faces many questions over new A.I. technologies and the people who build them.
The recent dispute follows a pattern of dismissals and dueling claims of wrongdoing among the A.I. researchers at the company. A.I. is one of humankind's most important endeavors, according to the chief executive of the parent company.
A group of researchers built a system that learned to recognize cats in videos on the internet. The expansion of the lab was a result of the executives being so taken with the idea that machines could learn skills on their own. The research group became a symbol of the company's ambitions.
Before she was fired, Dr. Gebru was seeking permission to publish a research paper about how A.I.-based language systems, including technology built by Google, may end up using the biased and hate-filled language they learn from text in books and websites. Dr. Gebru said that she had grown frustrated with the response of the internet giant.
Margaret Mitchell, the other head of the team, was fired by the company a few months later. The company said that Dr. Mitchell had violated its code of conduct.
The paper in Nature promoted a technology called reinforcement learning, which it said could improve the design of computer chips. The technology was hailed as a breakthrough for artificial intelligence and an improvement to existing approaches to chip design. The technique was used by the company to develop its own chips.
The machine learning technique was being applied to chip design for a long time, and it was published a year ago. The people familiar with the matter said that around that time, Dr. Chatterjee was asked if the approach could be sold or licensed to a chip design company.
Three people said that Dr. Chatterjee questioned whether the technology had been rigorously tested in an internal email about some of the paper's claims.
The debate about that research continued, and that's when Google pitched another paper to Nature. The names of two authors who had worked with Dr. Chatterjee were removed from the submission because of concerns about the paper's main claims.
The newer paper surprised some people. The people said that it had not followed a publishing approval process that Jeff Dean, the company's senior vice president who oversees most of its A.I. efforts, said was necessary in the aftermath of Dr. Gebru's firing.
The changes from the earlier paper did not require the full approval of the authors. A paper that challenged some of its claims was written by Dr. Chatterjee and a group of internal and external researchers.
The rebuttal paper was submitted to the resolution committee. The paper was rejected months later.
The researchers who worked on the rebuttal paper wanted to escalate the issue to Mr. Pichai and the board of directors. They argued that the rebuttal was not published because it did not uphold high standards of scientific excellence. The people said that Dr. Chatterjee was no longer an employee.
Ms. Goldie said that Dr. Chatterjee asked to manage their project in 2019. She said that when he criticized it, he could not substantiate his complaints and ignored the evidence they presented in response.
Ms. Goldie said that Sat Chatterjee had waged a campaign of misinformation against her and Azalia for over two years.
She said that the work had been reviewed by Nature, one of the most prestigious scientific publications. She said that the new chips were being used in the computer data centers.
The lawyer for Dr. Chatterjee said it was disappointing that the authors of the Nature paper were trying to shut down scientific discussion by defaming him.
Jeff Dean's actions to suppress the release of all relevant experimental data, not just data that supports his favored hypothesis, should be deeply troubling to both the scientific community and the broader community that consumes Google services and products.
Dr. Dean didn't reply to the request for comment.
After the rebuttal paper was shared with academics and other experts outside of Google, the controversy spread throughout the global community of researchers who specialize in chip design.
Some experts are unsure what the implications are of the research done by the search engine giant.
If this is working well, it would be a great thing, said Jens Lienig, a professor at the Dresden University of Technology in Germany.
All Rights Reserved © 2024
