Google wanted AI researchers to 'strike a positive tone' on the company
What you need to know
- Google began intervening in research papers by asking researchers to "strike a positive tone" when referring to the company's work.
- Former Ethical AI head, Timnit Gebru, was fired for her paper criticizing Google's language models.
- Gebru's paper has since been accepted by an outside scientific group.
Amid Google's already troubling legal battles around how it handles Search and ads, Google still faces internal struggles on its handling of former Ethical AI co-head, Timnit Gebru's departure. After presenting her research paper on the concerns of largely unchecked AI language models, such as those used in Google's search tools and the best Android phones, the company dismissed her paper saying it didn't account for new efforts Google was making to address these concerns, and therefore dismissed Gebru. That's putting it plainly, as the situation has presented itself a bit more complicated, but some information from Reuters shines a brighter light on how Google approached the paper.
According to internal policy changes that went into effect in June, Google began requiring its research staff to consult with legal and PR teams before pursuing certain topics that could be deemed as sensitive. These topics include race, gender, political affiliation, and face analysis, with the policy stating that these "seemingly inoffensive projects raise ethical, reputational, regulatory or legal issues."
Adding to this, researchers were asked that they "take great care to strike a positive tone." According to employees at Google Research, papers that were openly critical about Google's practices faced a much higher level of scrutiny than those that were not.
"It's being framed as the AI optimists and the people really building the stuff [versus] the rest of us negative Nellies, raining on their parade," said one outside AI expert. This is where Gebru's paper comes in.
Since 2016, the tech industry had grown more concerned over the ethical practices surrounding how AI is used. Google created the Ethical AI group in 2018, around the same time that it recruited Gebru with the promise of total academic freedom. Already pretty well regarded in her field, she took the position and, with the help of University of Washington linguist, Emily M. Bender, presented a draft on the paper that later led to her dismissal.
Google maintains that it did not fire her, but "accepted her resignation." But the outcry from Google's employees, as well as U.S. Senators, ask for full details on an internal investigation on the matter. Gebru's paper has since been accepted by the Conference on Fairness, Accountability, and Transparency.
Get the Android Central Newsletter
Instant access to breaking news, the hottest reviews, great deals and helpful tips.
Derrek is a long-time Nokia and LG fanboy who loves astronomy, videography, and sci-fi movies. When he's not working, he's most likely working out or smoldering at the camera.
I guess the bottom line for me here is that producing an accurate AI Ethics paper which clearly identifies shortcomings and issues with Google's AI does not necessarily seem to be mutually exclusive with presenting the company as a whole "in a positive light", which is what any company should want their employees to do. I must be missing something ... I am sure there is much that I don't know about that situation. But just from this article, it seems like folks seems to be conflating the Ethics of AI with the ethics on working for a company. Why it would be unreasonable for a company to require employees to not publicly trash their own company is a bit beyond me I guess.
It may be true, as the headline reads that "That's not how ethics works, Google." But I think it is also true that "That is not how the employer/employee relationship works, Gebru." The reporting I have seen indicates that Gebru submitted a list of demands for her continued employment, some of which appeared unreasonable to me. If an employee did that to me, I would let them leave as well. If she is going to continue taking a paycheck from Google, I don't think it is unreasonable to expect her to not publicly trash the company. She can do that from outside the company without making any compromises, so perhaps it is best that she leaves. Industry funded studies are often (and rightly) viewed with skepticism. Perhaps she will find a role in academia that will allow her to continue her work.