What Really Happened When Google Ousted Timnit Gebru | Hacker News Reader

What Really Happened When Google Ousted Timnit Gebru

wired.com

34 pointspcaversaccio9 comments

This article leaves so many questions unanswered. The author wrote many words but didn’t seem to research the subject well.

> Nor is it clear why her resistance—predictable as it was—prompted a snap decision to eject her,

Filled with many quotes like this where Gebru’s “resistance” was not predictable as I don’t think anyone expected her to react the way she did. In that, if this happened to 10 people, I don’t think 5 would send out unprofessional listserv emails .

> Dean repeated the point so often inside Google that some researchers joked that “I have an objection to Parrots section three” would be inscribed on his tombstone. The complaint made little sense to many AI researchers, who knew that grumbles about citations typically end with authors revising a paper, not getting terminated.

I don’t think Jeff Dean’s email [0] was concerned with missing citations, but that the work did not address and reflect the work in those citations that were contrary to what the paper reported.

The article does not understand or mischaracterizes the issues that are important to understanding this topic.

[0] https://www.platformer.news/p/the-withering-email-that-got-a...

> I don’t think Jeff Dean’s email [0] was concerned with missing citations, but that the work did not address and reflect the work in those citations that were contrary to what the paper reported.

Wouldn't that make the paper a PR piece for other projects at Google? Including some that may never be deployed?

When this story first broke, I commented elsewhere on HN that Dr. Gebru's first mistake was expecting Google to be in any way actually interested in her worthwhile mission. Of course it doesn't care about her, or sexism in the workplace, or the ethics of AI. It cares about making money and doing things that help it make more.

Google doesn't care if Dr. Gebru starts arguments or is rude or is friendly or is right or is wrong. It only cares that she doesn't screw with their money, and instead does things which helps them make more - in her case, by being a diverse face for their "ethical AI" idea, which is a facially absurd thing to expect an ad company to actually care about.

If she'd wanted to keep her job, the best thing to do would have been to focus on papers that critique the competition, and leave the critique of Google to people who don't work at Google. I don't blame her or anyone for making the mistake of thinking Google was sincere in its desire for her to do the job she did - but I think the truth is that she was fired not because she was Black or a woman or anything else (not that those things helped) but because whether she realized it or not, she wasn't doing the job Google actually wanted her to do.

This is pretty much exactly how I ended up thinking after almost a decade in industry. Anyone who thinks that social missions can be worked on inside a for-profit corporation is incredibly naïve. It is a difficult task even in non-profits, government institutions and politics; in essence, the systems are there for the systems' sake. The people within the system each have different motivations and reasons for being there. Even before joining a company, expectations of pushback should be the first thing on your mind if you are working in a potential social minefield like ethics governance.

Corporations exist to make money, period. Never expect social contributions to progress, worthwhile and long-term employment and a guaranteed paycheck or pension. Fighting for a worthwhile cause is always good, but if you're doing it on company time, expect that there's a good chance that you won't be at the company for long.

This. And I would also add how naive it is to think that you don't have to comply with the list of requests sent to you by the person giving you a paycheck each month.

Let be known, if you're accepting a salary from any person, company, or government, you are fully expected to do as they ask.

If you don't like it, don't accept the money.

If that were always true, whistleblowing would not be a thing.

I mean within reason obviously. Obviously your employer can't force you to get up and do jumping jacks at 3am for no reason either.

But as it pertains to work, provided it's not illegal, yea they can pretty much force your hand on a lot of things.

What was the profit motive behind hiring her?

I’m not challenging; I’m genuinely interested and baffled by these moves. I’d like to understand that chain of logic that ends in money.

I can't speak for Google, but something like this would be my guess:

Google has an issue with public perception where people avoid its products because Google's AI is creepy. The idea that the Internet knows too much about you is definitely a Thing in the culture.

Also, Google knows that people perceive it as not diverse, and not good to minorities or other protected groups who work for it.

Google wants to be perceived as a serious company that takes its obligations to the world and to the computer science community (open source, AI research, etc.) seriously.

Finally, somebody like Dr. Gebru, who is genuinely smart and knowledgeable, can probably make Google's products better in the process (per the article, sounds like she did).

Google believes it will do more business by some reasonable amount (perhaps not huge, so long as it's more than Dr. Gebru's compensation) if it appears to be addressing these three things above, in the eyes of at least some of the people who care about those things, and even if it doesn't, the work Dr. Gebru actually does is work Google thinks will help it make money by making its products better.

Put another way: If Dr. Gebru is a smart, productive, diverse face for Google's AI, it is much easier to believe Google's AI is friendly and "ethical" than if it's some white business-guy with no credentials, at least amongst the constituencies Google would like to convince to support it, and Dr. Gebru has knowledge areas Google wants to exploit for profit, just like any employee. If Dr. Gebru is the face of this part of Google's business, more people will be willing to come and work for Google on it, and, presumably, to eventually buy or at least not badmouth Google's services.

So it's not a direct line to profit. It's public perception --> hiring ability --> profit, or public perception --> marketing --> profit, or making GMail's suggestions less sexist --> avoid scandal --> avoid loss of profit. But all of this pales in comparison to public critique of Google's products --> perception problems --> loss of profit, plus it probably hurts a bunch of peoples' feelings internally, so Dr. Gebru is fired.

Does that help clarify?

Reply on news.ycombinator.com