Wednesday, July 28
Shadow

Google researchers dismissal is a perfect storm for employee unrest — and a big test for CEO Sundar Pichai – CNBC

Alphabet CEO Sundar Pichai gestures during a session at the World Economic Forum annual meeting in Davos.

Fabrice Coffrini | AFP | Getty Images

Google’s firing of renowned researcher Timnit Gebru last week touched some of its most troublesome workplace issues, creating a perfect storm for leadership.

It will be the biggest test of how CEO Sundar Pichai deals with employee unrest since founders Larry Page and Sergey Brin left the company about a year ago.

Gebru, a well-known artificial intelligence researcher and technical co-lead of Google’s Ethical AI team, tweeted Wednesday night that Google fired her over a disagreement about a research paper that scrutinized bias in artificial intelligence.

Gebru claimed the treatment was indicative of a broader pattern at Google. Although she had received numerous technical accolades, she’s also known for speaking up within Google about its lack of diversity and treatment of minority workers.

The company has offered little explanation, pointing only to AI chief Jeff Dean’s comments about the company’s processes for approving research papers.

The incident represents the intersection of four issues that have spurred employee unrest at Google in recent years: ethics in artificial intelligence, treatment of women and race, and the shift away from Google’s formerly open and free-wheeling culture.

Timnit Gebru in 2018.

Kimberly White | Getty Images

As support pours in for Gebru, it could hurt Google’s ability to recruit and retain the best artificial intelligence researchers, who have long been promised a top-notch working environment at Google. Employees from competitors have already begun offering Google’s AI employees a chance to learn more about their respective employers, should they decide to move away from the search giant.

Ethics in artificial intelligence

Because artificial intelligence is relatively new and potentially powerful, many in the tech industry are still grappling with how it should — and shouldn’t — be used.

This issue has erupted several times among Google’s tech-savvy and outspoken workforce.

In 2018, thousands of Google employees signed a letter urging their CEO to pull out of a satellite image-based defense contract called Project Maven, and about a dozen staffers resigned in protest. The protest came after employees grew concerned that the Defense Department would use Google’s AI research and expertise to kill people during warfare. Google eventually scrapped the deal.

In 2019, more than 1,400 employees signed a petition demanding Google not bid on a contract with the U.S. Customs and Border Protection or U.S. Immigration and Customs Enforcement, citing concerns over the Trump administration’s zero tolerance tactics against immigrants.

During this period, Google published a series of principles for developing AI that included guidelines such as “Be socially beneficial” and “Avoid creating or reinforcing unfair bias.”

This latter area was where Gebru concentrated her research. She’s most famously known for her work on a project called Gender Shades that revealed facial recognition technology was highly inaccurate for Black people. Her work contributed to IBM and Microsoft pausing sales of their technologies.

The research paper, which Dean asked her to retract, questioned whether building “large language models” — a type of AI that generates intelligible text — disproportionately benefits privileged people in society, and whether it might hurt marginalized communities, according to MIT Technology Review. It also discusses the environmental footprint of advancing AI.

Dean said he asked for a retraction because the paper “ignored too much relevant research” about large language models and did not take into account recent research into mitigation of bias in language models.

But Gebru’s colleagues quickly sounded off, stating Dean’s explanations didn’t line up with their experiences. A group of current and former Google employees went further on Monday, detailing a list of push-backs against Dean’s claims.

Treatment of women

Google is fresh off a landmark sexual harassment settlement, which came after employees and shareholders questioned how it handled claims of sexual misconduct and harassment, and more generally how it treated female employees. 

The settlement, which included more than 70 updates to internal policies, came one year after shareholders filed a lawsuit after Google quietly paid executives who were found to have engaged sexual misconduct. The company had structural gaps that executives with power were allegedly able to exploit or bypass, according to the filings.

In an email to employees after the settlement, Pichai encouraged them to read the highlights of the settlement, saying “I hope these commitments will serve as a strong signal to all of you that we are not going back in time.” 

Race

Gebru was one of Google’s few Black female leaders and has also been a vocal critic of tech and Google’s treatment of minority employees.

“We just had a Black research all-hands with such an emotional show of exasperation,” Gebru said in an email to colleagues that criticized the company’s handling of diversity and inclusion efforts and was sent a few days before she was fired. “There is zero accountability.”

While Google has touted its diversity hiring initiative, it has made little progress in building and maintaining a diverse workforce. Attrition rates for Google’s Black U.S. employees remain stagnant from a year earlier at 12% — higher than the average Google employee — with Black females seeing a particular spike in attrition from last year, up 18%, according to the company’s 2020 diversity report.

After the George Floyd protests in summer 2020, Pichai promised to address Black employees’ concerns, adding that the company would commit $175 million toward supporting Black businesses and increase “underrepresented” people in its leadership by 30% by 2025. Executives including Dean publicly promised to stand against discriminatory treatment, which can be subtle.

After Gebru’s departure, employees resurfaced some of Dean’s tweets from July: “When the bad behavior is disproportionately exhibited by those who are white and the people to whom it is directed are not, we need to call this what it is: racism.” Colleagues alleged Dean’s explanation for Gebru’s paper rejection was a result of a “standard which was applied unevenly and discriminatorily.”

In September, CNBC reported that employees of Google’s recruiting arm had requested more clarity and transparency when it comes to the company’s June commitments and its retention plans. Employees and recruiters said they were worried minority turnover would continue without clearer diversity goals.

“It is crucial that we are working towards building a culture of inclusion,” the leaked letter said. “We don’t want to bring in talented students who end up leaving because the future isn’t what they thought it would be.”

Transparency and retaliation 

As the company remains silent on its handling of Gebru’s employment, it has left room for conflicting interpretations of what happened. More than 2,000 Google employees, managers and industry supporters have petitioned Google for answers.

The silence exposes a familiar gap between leadership and employees as the company has begun shifting from a once-open, idealistic and free-thinking culture to a more traditional corporation. Since then, the company has begun trying to strike a balance between pleasing its employees and cracking down on heated debate.

Some employees pointed to the irony that Gebru’s tweets came on the same day The U.S. National Labor Relations Board filed a complaint against Google alleging the company retaliated against and surveilled employees while blocking workers from sharing work grievances and information with each other using general tools like calendars, email, meeting rooms, and an internal communication tool at Google called MemeGen

The company again cracked down on its internal messaging boards in September as its internal data showed discussions getting more heated as employees remain working from home. Google’s internal community management team said it saw a rise in posts flagged for racism and abuse.

Google employees told CNBC that 2019 was a pivotal year for the company and that some had left, partly due to a lack of transparency from management. Workers said the company’s culture has turned into the opposite of what the founders had long desired.

“Executives really engaged with the debate and the dialogue which I don’t think you see now,” said Claire Stapleton, a 12-year veteran who led a push for policy changes, including the walkout. She left Google last summer, claiming retaliation by company leaders.

Watch Now: Renowned AI researcher says Google abruptly fired her, spurring industrywide criticism

Leave a Reply

Your email address will not be published. Required fields are marked *