Little did Luke Stark look for money on Google in November that he would turn down $ 60,000 from the tech giant in March.
Stark, Assistant Professor at Western University in Ontario, Canada, studies the social and ethical implications of artificial intelligence. At the end of November he applied for one Google Research Scholar Award, a non-binding research grant of up to $ 60,000 to support professors early in their careers.
He applied for the award and said, “Based on my feeling then that Google is building a really strong, potentially industry-leading team in ethical AI.”
Soon this feeling began to dissipate. In early December, Timnit Gebru, co-head of Google’s ethical AI team and a prominent black woman in a predominantly white, male area, abruptly left Google. On Wednesday, December 2nd, she has tweeted that she was “fired immediately” for an email she sent to an internal mailing list. In the email, she expressed dismay at the continuing lack of diversity in the company and frustration with an internal process related to the review of a then-unpublished one research paper on the risks of building bigger and bigger AI language models – a vivid type of AI increasingly important to Google’s huge search business.
At the time, Gebru said Google AI leadership had told her to withdraw the paper from exam for a presentation at a conference, or remove her name from it. Google accepted Gebru’s resignation over a list of demands she emailed that needed to be met in order for her to continue working at the company.
Gebru’s fall sparked months of crisis for the company, including Employee departures, a change in leadershipand an apology from the Google CEO for some employees questioning their place there due to the circumstances surrounding Gebru’s departure. Google ran one internal investigation into the matter Results of this have been announced On the same day, the company fired Gebru’s co-team leader Margaret Mitchell, who had consistently criticized the company on Twitter after Gebru left. (Google cited “multiple violations” of its code of conduct.) Meanwhile, researchers outside of Google, particularly in the AI space, are increasingly suspicious of the company’s historically prestigious grant and angry with Gebru and Mitchell’s treatment.
All of this became sharp for Stark on Wednesday March 10th when Google sent him a congratulatory letter in which he received $ 60,000 for his proposal for a research project examining how companies are adopting AI that uses emotions be recognized. Stark said he immediately felt he had to turn down the award to show his support for Gebru and Mitchell, as well as those who are still on Google’s ethical AI team.
“My first thought was, ‘I have to refuse,” Stark told CNN Business.
Stark is among a growing number of scholars who cite Gebru and Mitchell’s conclusions about recent decisions regarding the company’s loss of funding or opportunity. Some AI conference organizers are considering sponsoring Google. And at least one academic who has received a large check from Google in the past has now stated so will not seek his financial support until changes are made in the company.
“I can no longer accept funding in good conscience from a company that treats its employees this way,” Vijay Chidambaram, an assistant professor studying storage systems at the University of Texas at Austin, told CNN Business. Chidambaram received $ 30,000 from Google for a research project in 2018.
The money involved is of little importance to Google. However, the growing impact of tensions between Google and its ethical AI team poses a risk to the company Reputation and stature in the AI community. This is vital when Google is competing for talent – both as employees of the company and names associated with it in the academic community.
“I think this is more common than even the company realizes,” said Stark.
Lose weight in solidarity
Despite his initial inclination, Stark did not immediately turn down Google’s award. He spoke to colleagues about what he was up to – “People supported every decision I made,” he said – before sending his reply to Google the following Friday. He thanked the company for the “vote of confidence” in its research, but wrote that “he received this award out of solidarity with Dr. Gebru and Mitchell, their teammates and anyone else who found themselves in similar situations, ”CNN Business said in an email.
“I look forward to the opportunity to work with Google Research again when the organization and its leaders have reflected on their decision on this case, addressed the harm they have caused, and committed word and deed to promoting critical research and products that do Support justice and justice, ”wrote Stark.
He tweeted on his decision to decline the award and make it public, noting that many people cannot afford to decline such funds from Google or other companies. Stark can do without the money because his department at Western University is adequately funded. Google’s award would have provided additional research funding, he said.
“All we can do is what we can reasonably do – and that was something I thought I could,” said Stark tweeted.
Gebru said she appreciated Stark’s action.
“It’s a pretty big deal for someone to turn down Google sponsorship,” she told CNN Business. “Especially someone early in their career.”
A Google spokesperson said the company has awarded more than 6,500 academic and research grants to people outside of Google in the past 15 years. According to the spokesman, Stark is the first person to turn down one.
“It was a real fiasco the way they were treated”
But Stark’s decision is just the latest token of solidarity with Gebru and Mitchell.
The first obvious sign of anger came shortly after Gebru left Google. ON Medium post that decodes their departure and the demand for transparency about Google’s decision on the research paper quickly received signatures from Google employees and supporters in the academic and AI fields. By the end of March, the number of supporters had grown to nearly 2,700 Google employees and over 4,300 others.
At the beginning of March, the conference to which Gebru and her co-authors had submitted the paper had the ACM Conference on Fairness, Accountability and Transparencyor FAccT has discontinued its sponsorship agreement with Google. Gebru is one of the founders of the conference and a member of the first FAccT Executive Committee. Google has been a sponsor every year since the annual conference began in 2018. Michael Ekstrand, co-chair of the ACM FAccT network, confirmed to CNN Business that the sponsorship has ended and said the move was “in the best interests of the community” and that the group is “rethinking” its sponsorship policy for 2022 becomes. Ekstrand said Gebru was not involved in the decision.
Also in March, two scientists protested against Google’s actions by tweet The They decided not to attend an online robotics research event that was by invitation only. Hadas Kress-Gazit, a Cornell robotics professor, was one of them; She said she was invited in January but grew more reluctant as the event drew near.
“By the way, it was a real fiasco [Gebru and Mitchell] were treated. Nobody has apologized to them yet, ”she said in a recent interview with CNN Business. “I don’t want to interact with companies that behave this way towards top researchers.”
Google’s efforts to push the boundaries in AI
Google is aware that its reputation as a research institution has been damaged in the past few months, and the company has announced that it will fix the problem. At a recent city hall meeting held by Google, first reported on by Reuters and audio received from CNN Business, the company outlined changes to its internal research and publication practices.
“I think the way to regain confidence is to keep publishing cutting-edge work in many, many areas, including pushing the boundaries on topics related to responsible AI and publishing things that are profound to the research community are interesting from the best opportunities to continue to be a leader in research, ”said Jeff Dean, head of AI at Google. He answered a staff question about outside researchers that they would read Google articles “with greater skepticism” now.
Gebru hopes that, as with FAccT, more conferences will reassess their relationships with tech company research laboratories. In the past, much of the work on developing and studying AI was done in an academic setting. However, as companies find more commercial uses for the technology, the lines between the academic and corporate worlds are blurring. Google is just one of many tech companies that has had a huge impact on academic conferences where many of its researchers’ articles are published. The staff sit on conference boards and sponsor numerous conferences each year, sometimes for tens of thousands of dollars.
For example, Google and some subsidiaries of parent company Alphabet were listed as sponsors of the International Machine Learning Conference (ICML) and the Conference on Neural Information Processing Systems (NeurIPS) with platinum and gold awards of $ 20,000. in 2020 – both major AI conferences. Some of the company’s employees sit on their organizing committees.
ICML President John Langford said the conference was “currently open to sponsorship” from Google for the 2021 conference scheduled for July.
“There is much discussion about how ICML as a conference should promote good machine learning culture and practice, with future sponsorship policy being part of that discussion,” he added.
NeurIPS executive director Mary Ellen Perry said the conference has not yet called for sponsorship annually, but requests will be “assessed against a set of selection guidelines set by this year’s sponsorship chairs.” NeurIPS is planned for December.
However, for Stark and other members of the academic research community, the criteria for accepting funds from Google have already changed.
“Extra research funding would be great,” said Stark. “But it was something that I felt I just couldn’t take.”