Google researcher claims fired on paper, highlighting bias in AI

A renowned Google researcher said the company fired after criticizing the approach to employing minorities and the biases embedded in today’s artificial intelligence systems.

Timnit Gebru, co – leader of the Google Ethical AI team, beep Wednesday night that he was fired for an email he had sent a day earlier to a group that had company employees.

In an email revised by The New York Times, he expressed outrage at Google’s response to his and other employees ’efforts to increase employment for minorities and call for artificial intelligence bias.

“Your life gets worse when you start standing up for underrepresented people. It starts to annoy other drivers, ”reads the email. “There’s no way more documents or more conversations can achieve anything.”

His departure from Google highlights the growing tension between Google’s outspoken workforce and buttoned top management, while expressing concern about the company’s efforts to build fair and reliable technology. This could have a cooling effect on both black technology workers and researchers who have left the academy for high-paying jobs in Silicon Valley in recent years.

“His dismissal only indicates that scientists, activists and scientists who want to work in this field – and black women – are reluctant to see him in Silicon Valley,” said Mutale Nkonde of Stanford Digital Civil Society Lab. – Very disappointing.

A Google spokesman declined to comment. In an email to Google staff, Jeff Dean, who oversees the work of Google’s artificial intelligence, including Dr. Gebru and his team, called his departure “a difficult moment, especially given the important research topics he was involved in and how deeply we care about responsible artificial artificial intelligence research as an organization and as a company. “

After years of an ubiquitous environment where employees had free-flowing discussions in corporate meetings and online message boards, Google began to push back on workplace speech. Many Google employees looked at the new restrictions, arguing that the company had broken away from a tradition of transparency and free debate.

On Wednesday, the National Labor Relations Board said Google allegedly violated labor law when it fired two employees who were involved in workforce organization. According to a federal agency, Google illegally inspected employees before they were fired.

Google’s battles with workers who have talked about dealing with the company’s sexual harassment in recent years, as well as their work with the Department of Defense and federal border agencies, have eroded its reputation as a utopia of technical workers, with high pay, benefits and leave. .

Like other technology companies, Google has been criticized for not doing enough to address the lack of women and racial minorities in its ranks.

The problems of racial inequality, especially the mistreatment of black employees in technology companies, have plagued Silicon Valley for years. Coinbase, the cryptocurrency’s most valuable startup, has seen the emigration of black employees over the past two years, according to workers, in racist and discriminatory treatment.

Researchers are concerned that people who build artificial intelligence systems may incorporate their own bias into the technology. Numerous public experiments in recent years have shown that systems often interact with people in color in different ways – perhaps because they are underrepresented among the developers who create the systems.

Dr. Gebru (37) was born and raised in Ethiopia. In 2018, as a researcher at Stanford University, he helped write an article that is widely seen as a turning point in efforts to accurately define and eliminate the bias of artificial intelligence. He joined Google back that year and helped build the Ethical AI team.

After hiring researchers like Dr. Gebru, Google has painted itself as a company dedicated to “ethical” commitment to artificial intelligence. But he is often reluctant to disclose the flaws in his own system.

In an interview with The Times, Dr. Gebru said his frustration stemmed from the company handling a research article he had written with six other researchers, four of them at Google. The article, also reviewed by The Times, highlighted the flaws in a new kind of language technology, including a system built by Google that supports the company’s search engine.

These systems learn the whims of language by analyzing vast amounts of text, including the analysis of thousands of books, Wikipedia entries, and other online documents. Because this text contains biased and sometimes hateful language, technology can result in biased and hate speech.

After he and other researchers submitted the article to a scientific conference, Dr. Gebru said a Google manager demanded that he either withdraw the article about the conference or remove his name and the names of other Google employees. He declined to do so without further discussion, and in an email sent Tuesday night, he said he would resign after a reasonable amount of time if the company couldn’t explain why he wanted to withdraw the paper and respond to other concerns.

The company’s response to the email said it could not meet its needs and its resignation would be accepted immediately. Your access to corporate email and other services was immediately revoked.

In a letter to employees, Mr. Dean said Google is respecting its “resignation decision”. Mr. Dean also said that the article does not acknowledge recent research that shows ways to reduce the bias of such systems.

“It was inhumane,” Dr. Gebru said. “There may be a reason to stop our research. But the most annoying thing is that they are unwilling to have discussions about why. “

Dr. Gebru’s departure from Google comes at a time when AI technology is playing a larger role in almost every area of ​​Google’s business. The company has tied its future to artificial intelligence – whether with its voice-based digital assistant or automated placement of marketers ’advertising – as a breakthrough technology that will make the next generation of services and devices more intelligent and capable.

Sundar Pichai, CEO of Alphabet, Google’s parent company, compared the emergence of artificial intelligence to electricity or fire, stating that it is essential to the future of the company and computing. Earlier this year, Mr Pichai called for more regulation and responsible management of artificial intelligence, arguing that society needs to balance potential damage with new opportunities.

Google has repeatedly committed itself to eliminating the distortions inherent in its systems. The trouble, Dr. Gebru says, is that most of the people who make the final decisions are men. “Not only are they saying not to prioritize more people from minority communities, they are silencing their voices,” he said.

Julien Cornebise, an honorary associate professor at University College London and a former researcher at DeepMind, a promotional artificial laboratory owned by one of Google’s parent companies, was among several artificial intelligence researchers who say Dr. Gebru’s departure reflects a bigger problem in the industry.

“This shows that some large tech companies only support ethics and fairness, as well as other reasons for artificial intelligence in the social sense, as long as their positive PR impact outweighs the extra scrutiny they cause,” he said. “Timnit is a brilliant researcher. We need more than he does in our field. “