AI researcher Timnit Gebru resigns from Google • Hacker News Board

Re: AI researcher Timnit Gebru resigns from Google

#921

Post by blaser-waffle » Mon, Dec 07, 2020, 2:45 PM UTC

LordFast wrote: post #736

jsmith45 wrote: post #486

Earlier quoted context omitted.

It sounds to me like Google has a policy that papers must be internally reviewed before they can be "published". She understood publish in the academic sense, while Google views sending the paper out for conference review as publishing. The paper ends up failing internal review, so per policy it must be promptly retracted. This is confusing to the academic who expects to get access to the raw review responses so that…

Me thinks it's a common misconception for strongly opinionated people to believe that companies want to hire them in order for them to "fix things and make the company a better place". In reality, what every company/team/management wants is someone who will give them some constructive criticism on small mistakes here and there that are easy to fix, but ultimately act as a supportive shield for the company against /ot…

Quite jaded but not wrong per se.

Usually is a reason they hire consultants and contractors to do say the unpleasant truths. They can be ignored, reframed, bribed to say whatever, and have the luxury of saying ugly things and then fucking off not long after.

Re: AI researcher Timnit Gebru resigns from Google

#922

Post by uoaei » Mon, Dec 07, 2020, 8:06 PM UTC

Copenjin wrote: post #775

uoaei wrote: post #468

Earlier quoted context omitted.

This is a very dangerous stance to take. You are advocating for a purely authoritarian mode of management, applied universally without concern for context.

Not saying that it's desirable or that's not horrible, but isn't that the default mode? Or at least that's the world I've lived in, if you are an employee, in the end your opinions don't actually count, you are just a cog. Your position in the hierarchy is what matter. And that's why at some point you need labor unions (not saying high paid tech workers needs them) and at some other point you start considering stop b…

Resigning to that power relation is defeatism, and unhelpful for the maintenance of labor rights.

Re: AI researcher Timnit Gebru resigns from Google

#923

Post by tdeck » Tue, Dec 08, 2020, 6:25 AM UTC

beaunative wrote: post #886

tdeck wrote: post #731

Earlier quoted context omitted.

In any large company there are lots of process rules and internal deadlines that are unrealistic for whatever reason and don't get followed in practice, leading to many last-minute exceptions. I don't know about the specific case of research papers, but if would not surprise me at all if "one day before the deadline" were a common submission timeframe at Google. Of course that not ideal and probably sucks for the tea…

A lot of process rules are unrealistic, but that doesn't mean this one is, and that can't be used to conclude that all internal processes are unrealistic and wouldn't be followed. It's an quite unrealistic statement to make. Most places rely on internal processes to work, and it would be extremely rare for any case of exception.

> Most places rely on internal processes to work, and it would be extremely rare for any case of exception.

That's a pretty broad statement with no supporting evidence. In fact, I'd argue that most large organizations rely on internal processes to be ignored or bypassed just as often as they rely on other processes to be followed.

I've been on teams that required "every feature" to go through a design review. Guess what? Suddenly a lot of what we were already working on stopped being "features" and started being "small bug fixes". Approvals for launches are often sought from the last few approvers just before launch because the list of required approvers is so long, and the VP of engineering isn't going to approve until everyone else has already approved etc... This kind of shit is par for the course at a large company because the alternative is basically a "work to rule" strike.

Re: AI researcher Timnit Gebru resigns from Google

#924

Post by gcblkjaidfj » Wed, Dec 09, 2020, 7:31 PM UTC

beervirus wrote: post #842

sandworm101 wrote: post #179

Earlier quoted context omitted.

One paragraph has four parentheticals. I wonder which humanities background, other than Law, finds that acceptable.

I'm a lawyer, and I would never write in such an intentionally obtuse style.

Saying that legalese is not intentionally obtuse is the real flex :)

Re: AI researcher Timnit Gebru resigns from Google

#925

Post by kazamaloo » Thu, Dec 10, 2020, 4:55 AM UTC

mehrdada wrote: post #798

kazamaloo wrote: post #791

Earlier quoted context omitted.

She sends for approval 1 day before the conference deadline, proceeds to submit the paper with conditional approval, and waits for the actual approvals. It is common practice with one side effect: if reviewers don't like certain parts of the paper, they can ask you to withdraw it (since you don't have an option to update the paper at this stage). If the paper was submitted for approval before submitting to the confer…

Nowhere in the account of the story from the other party anything is specified as "conditional" nor whoever issued the initial approval is mentioned at all. More weirdly, in her account of the story, the "back-and-forth" came from, tada... HR department, which is super weird. Look, I am not necessarily picking her side in the more general story, but it is apparent this paper submission story in isolation is super wei…

I agree with you for almost everything you said. This case is definitely unusual. What I mean by conditional is that the person who approved it initially is not an actual reviewer.

For your last sentence, I think Jeff uses that 2-week policy to state that his position is "technically" right.

Re: AI researcher Timnit Gebru resigns from Google

#926

Post by AI_researcher » Thu, Dec 17, 2020, 7:40 PM UTC

anonvenger wrote: post #680

wonnage wrote: post #632

Earlier quoted context omitted.

And this is what HN has devolved to, calling a leading Ethical AI researcher a douche!

While there is some overlap, generally speaking Identity Politics != Ethics. She's an Identity Politics researcher, narrowly focused on race and gender topics. Not some philosopher or religious figure interested in the flourishing of humanity at large. Here is an abstract from a larger work, presumably her PhD thesis: https://arxiv.org/pdf/1908.06165.pdf . Critical gender and race theory through and through.

Amen. Her paper is citing dated problems in AI models. These problems have been known for at least 4-5 years, and the responsible AI community has been working on it. No progress mentioned. The paper read like an editorial rather than actual AI research.