AI researcher Timnit Gebru resigns from Google

Unless Google deliberately changed the enforcement of the policy just to mess with her, she should have known the policy. It doesn't seem to be a complicated process, and 2 weeks is a reasonably short time to wait.

On the other side, Google has been in this game long enough, that they must know a paper can be updated in this case. So there wouldn't be a misunderstanding there, either.

What exactly happened is unclear so far. But I doubt it will come down to just a series of misunderstanding.

On a side note, I suspect that working research for a company with a conflicting interest in the same area is generally a bad idea. At some point, the dilemma of either leaving or stopping caring about it would present itself.

Paper writing is painstaking and mentally tiring. It would be much more efficient if they reviewed results and an abstract and then decided based on that "Go ahead and write the paper and publish it" or "Don't waste time writing it up".

This feels like a classic privileged tantrum which has blown up in the face because Google refused to bend this time.

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 /other/ criticisms. The more credible that someone is, the better, and the more sophisticated that act of supportive shielding is, the better.

In exchange, the company provides support, financial or otherwise, for that someone's personal career ambitions in becoming rich and famous themselves.

Anyone else see it this way? Or am I just too jaded these days?

Asking for the review notes is one thing, it seems like the main demand Jeff refused was giving the names of every reviewer. Why would the names of the reviewers be relevant to the academic, unless they specifically wanted to use the reviewers identity in order to undermine the notes?

This is pure speculation, but it looks like there may have been some additional backchannel from the conference reviewers to their buddies at Google that caused a follow-on sensitivity or objection later from other functions at Google, namely PR & Policy (based on her tweets), which may have been why Timnit was super curious about the identity of the feedback authors.

There were clearly conflicts brewing below the lid. This was just taken as the final straw. On both sides, I'd reasonably expect. She was dissatisfied with a bunch of things that she couldn't makr public, so she boiled over on this one. That's why many people are surprised this is such a big deal.

And then she gave a ultimatum, which was great for both sides. For the employer since they could use it to get rid of her right away and make it look it's because of this (and "technically" they are right). And for her because she can use it for maximum outrage. It was a calculated provocation. Come on, if you really want to be part of a process of change, that's not how to achieve that. She is smart enough to understand this.

That seams a big jump. Isn't it more likely that the treatment is due to pushing for ethics in an organization whose business model is antithetical to ethics?

At most CS conferences, there is no second round of reviews (apart from a few PL ones that have been experimenting lately). If you get rejected, it's with the expectation that you make whatever changes you see fit and then submit anew to a different conference. In addition, there is generally no way to update a submitted paper after the submission deadline in such a way that the changes can still be taken into account in the single round of reviews, and making major changes after the accepting venue's review stage is very much frowned upon (as those changes would then not be reviewed). So her options would have been:

1. retract the paper;

2. get the paper rejected from the conference, and resubmit it later;

3. get the paper accepted and have it published without the major changes requested by Google, thus violating Google's policy/interests;

4. get the paper accepted and perform the major changes requested by Google at the camera-ready stage, thus violating academic convention and possibly even specific rules set out by the accepting conference.

2 is mostly 1 with more of everyone's time wasted, and 3 and 4 both violate some obligation or another. From Google's point of view, she might also well have gone for 3 without keeping them in the loop to present them with a fait accompli, and her enormous amount of social capital may have made it easy to get away with. Demanding 1 therefore seems like the most reasonable choice if they want to retain a credible norm of subjecting publications of people who work for them to prior review.

Out of curiosity do people think that a company either should or would publish self-funded research that's detrimental to their PR? I'm not sure that this is a realistic expectation with our current methods of company governance (which should probably be changed). I'm assuming the research isn't overly safety oriented and is more like "overuse of google leads to decreased overall well-being" rather that "use of the cigarettes that we sell leads to significantly reduced life expectancy". Of course that statement begs the question of where the line between the two lies.

“Let’s generate fake data with GANs to prevent our AI from discovering politically uncomfortable regularities!”

It’s ugly work but someone has to do it, right?

I wish this were true.

The Twitter replies read a lot like her Google message thing. She’s got a lot of good things to say but needs to work on the delivery.

I think she’s playing the victim and race card a lot more than she should. She’s got all the ammo she needs to make her arguments without the angry victim tone.

In this case, the issue is that "it's the dataset" is attacking a straw man: nobody is accusing ML practitioners to be explicit racists actively conspiring for their models to be racially biased. Everyone understands that it is (often) a result of the underlying dataset.

But, the argument goes, understanding the genesis of such bias isn't enough to excuse it. At least not when those models are put to actual use in, say, law-enforcement contexts. Or credit agencies, or hiring decisions, or really anything.

If your dog happened to bite any red-haired kid it came across, it doesn't matter that you think red-haired kids are just as good as others, or that the dog was once mistreated by a red-haired kid. You're not going to allow the dog to get near any red-haired kids. And if you do, you'll face charges of at least negligence when some red-haired kid gets mauled.

To tie this back to AI, it would mean that as long as your models produce racially biased results, they simply aren't ready for deployment or publication. Go find better data until your work is no longer liable to inflict harm on anybody.

That's the standard for all other industries: if your car is shown to reliably kill pedestrians below a certain height in crash tests, you go back to the drawing board. Adding a sticker "don't operate car in the vicinity of children" isn't enough.

You can disagree with that reasoning (although you'd be wrong). What you can't do, as a leader in AI, is to pretend to have never heard of it, or that it is so obviously wrong that it warrants no reply whatsoever.

> have to call out the "I'm so tired" bit every time I see it.

so... I guess you're tired of it?

Yeah, it sort of implies that, yes, you're angry about something, but actually have no constructive advice or solutions, other than to accuse someone of being a bad person or even a racist.

The "I'm so tired" bit is similar to the "I'm not here to teach you" bit. Maybe someone can't teach you because they actually have nothing to teach.

She says 'let us lead and you follow' -- sure...where? What are we concretely doing about the "in-built bias" problem?

Maybe someone could post it here when it's published?

We don't know the full list of demands, but Jeff Dean shared one demand that people may find unreasonable. Other demands may have been much more reasonable, but those weren't shared.

As such, I find it really quite baffling that someone would be unwilling to share that scientific feedback with her, or who raised it.

Where did you read that?

in the email shown,she told a listserv called "Women and Allies" to stop writing docs that are an attempt to better the situation in terms of diversity, equity, inclusion as writing those docs are like pissing in the wind in her opinion

It will be difficult for her lawyers to prove that she was let go as a retaliation, which it is not. The only thing which is apparent is that her employer did not value her work as much as she thought, so that her demands and her threat to resign (and the tarnishment of the Google brand because of the Twitter threads which she would write about it) would actually turn into a "clean" opportunity to let her go.

That is what happens when you are supposed to negotiate, and go the confrontational way. You have to estimate your leverage accurately before the full-blown war, otherwise it is a blitzkrieg.

> Timnit responded with an email requiring that a number of conditions be met in order for her to continue working at Google, including revealing the identities of every person who Megan and I had spoken to and consulted as part of the review of the paper and the exact feedback.

Isn't the word "mutiny"?

No idea if that's happened here, but as a manager I know what I'd do if an employee -- any employee -- told me to to meet their demands or they would quit.

[1] https://www.nbcnews.com/pop-culture/tv/muppets-studio-says-k...

I'm not applying this to Gebru vs Google, as in this specific case, we don't have enough evidence to know what exactly happened.

While I don't like the Twitter mob per se, the idea that a person within an institution cannot publicly criticize the institution is ridiculous. Rank-and-file members should be able to challenge the leaders. The employers can already fire employees for almost any arbitrary reason. Don't like the way a subordinate talks about you? Fire the person.

Colleges promote (or should promote) independent thinking. Independent thinking makes it necessary that employees might disagree with the employer. Is that good for the efficiency of the company? Hard to say. But to say the society must cultivate a culture of obeisance such that people can work for companies better is upside-down. Companies should be built to favor independent thinkers.

And yet, decades ago, when labour had real power and unions were a force in the economy, we had no problems with labour driving change within organizations.

I think what we're seeing now is what happens when individuals are disenfranchised from the political process and labour has no way to organize to drive broader change: individuals, frustrated with their lack of power, speak out and there is retaliation. And where there is a lack of organization, this anger has no focus and what you see is a chaotic lashing out.

I think the same is true of populism in general: when you render individuals powerless you prime them to be welcoming to the populist who promises to fight for them.

Is that an excuse for either Gebru or Google's behaviour, here, specifically? I have no idea. None of us have enough of the details to really understand what happened.

But the idea that labour shouldn't challenge leadership, or that the public in general shouldn't challenge the establishment, is a product of decades of union busting and concentration of power in the political class that we shouldn't accept just because it's become the status quo.

> I said here are the conditions. If you can meet them great I’ll take my name off this paper, if not then I can work on a last date.

To employers, that qualifies as a conditional resignation.

If you are instead just saying that Google should have let it slide because it was just an emotional outburst ("anger and frustrations") and not meant to be taken seriously, well, in what context can we hold adults to their word at all? Email is written communication, not an offhand verbal remark uttered in a tense meeting in anger. You should be able to see what you are about to send while typing it, and have ample opportunity to look over it again before you press that "Send" button. If your ability to review your words has atrophied, perhaps under the influence of social media which encourages "unfiltered" stream-of-consciousness venting, then this ought to be on you, simply because it is doubtful that society can function without some possibility of making binding statements and we do not have anything more commitmental than email widely deployed (and operable at the pace of the modern workplace) in work-from-home times.

Even pull the victim card, blast it all over social media. Sorry, that's a major disservice to everybody _actually_ being dehumanized. She's just dissatisfied with her employer and wants to make the biggest possible splash now.

If a company publishes only papers that clearly advance its interests, but the research is sound and independently reproducible, does that bias really lessen the perceived value and quality of the research?

The done thing is generally for the reviewers to suggest additional references to improve the paper. If any mention of related work is entirely missing, then that's another matter, but a reviewer thinking that the authors should cite a particular piece of work (e.g. the reviewer's own paper) is not normally something that could lead to outright rejection.

There is not a single software team on planet earth that would not want Jeff as their technical leader.

It is like comparing a high school athlete to Usain Bolt.

Industrial researchers and non-PIs (IE, research analyst working in a PI's lab) don't have the same freedom and integrity rights as a PI. To me this seems like a reasonable (and probably the only reasonable) policy for industry.

I don't really think it's dystopian- if a paper I wrote reached the point of needing a retraction it means some serious messup occurred or there's a major PR fiasco.

Note that in this case I am not sure if there was a retraction, or an desubmission- the paper hadn't been published yet so it's not really a candidate for retraction. The term for this is 'withdraw submission'

It's just as if you'd want to open source some code that you wrote for your job, and suddenly the boss says "hey, don't do that" - there are many reasons why they might want to not publish it, and it's not your code to publish as you want.

To some extent, you have to realize that one day you will fall on the wrong side on the fence and you'll be the victim of your own creation.

In this world, where there are bills to pay and you can't disadvantage yourself if you want to be ok. Having bills to pay is, I agree, just mad.

Google went through their standard process, and the paper was rejected.

If they have actually approved dozens of Timnit's papers before, then the likelihood that this rejection is due to discrimination seems low.

Big red flag. Why are the identities of reviewers important here? Did she plan to take those reviewers to the court of public opinions for a trial/exposure?

A lesson is delivered in time IMO. Those entitled people need to be called out.

They set themselves up for failure regardless of the paper's content.

If I say "I demand X or else I resign", (I don't know if that was actually said) I think it's fair to consider that a conditional resignation.

> A person: I think bias in machine learning is caused by bias in training data. > Me: Why are you ignoring all I've been saying? You're a racist. You're a white man who oppresses. I'm not going to talk to a bigot like you.

Repeat this a few times, the attacked will be punished both online and offline. Justice served.

Basically (to summarize a lot), his point was: an ML model is only as good as the data you feed it. If, say, the photos for your face recognition model are only of white men, then obviously the model will do well on white men, while (possibly) not doing as well on other races or genders. This is a statement of fact, and nothing controversial about it.

But she took offense to it, and started insulting him on Twitter. He got tired of defending himself, and just quit twitter.

His tweet doesn't rule out the possibility that he secretly has a malicious set of incentives, just like your handle doesn't rule out that you're a Communist spy sent to fracture the American psyche. Yet, its ridiculous. I feel like a lot of the world in 2020 is for some reason a hostile filter is applied ~99% of the time on the internet, and you _rarely_ see that in personal interactions in real life.

Coming back to the original tweet: if you changed the training data have more black people instead of white, would it perform the same but with inverted racial biases? Maybe? You really can't know without doing it. It might generate faces with a dark complexion but also big distortions or unrealistic colors. The original model doesn't just produce white people because of the training data, the hyperparameter tuning, perhaps the entire architecture, would have been modified until it produced acceptable outputs... using white training data. Ultimately the engineers and other humans behind the scenes are the arbiters of success, loss functions are chosen by humans, and swapping training data on the same model won't change those early decisions.

In another context I'm sure LeCun could have offered his somewhat reductionist take on this example of bias (something I might have done myself – reductionism flows in technologists' veins!) A discussion could have ensued, and everyone could have come out with a better understanding. A hot Twitter thread isn't where that will happen. Neither LeCun nor Timnit have the power to change what Twitter is. LeCun (reasonably!) doesn't like the nature of the discussion, and he leaves, and I think that's OK.

I don't think we can infer much from a few highly publicised events.

Of course social activists may at times cross the line and be unnecessarily rude or abrasive. But I'd have more sympathy because their fundamental job is to be subversive and critical of people.

The problem is that if you focus on race, gender and/or sexual orientation as an important factor underlying interactions, if you're not careful you can soon come to see race and gender as the primary driving force that dictates the outcomes of all interactions, and soon all issues get reduced down to a matter or racial, sexual, or class struggle.

The problem is the real world doesn't work that way, it is a complicated interacting network of events that are often only 2nd or 3rd order events from other completely different decisions made long ago.

To try and reduce all interactions and societal problems to a matter of a 1 or 2 factors is like trying to make money on the stock market solely by looking at the Fed's monetary policy. Sure the Fed's monetary policy is an important overall factor, and in some cases the most important factor in certain changes in the stock market, but that doesn't mean that there aren't a thousand other moving parts to focus on.

The problem is when we interpret every event through a single filter, or give one filter primacy in all interpretations of events that filter now colors everything we see.

So for example if my primary filter for understanding the world is race relations and/or critical race theory, any time someone disagrees with me I really do feel as though it is a racially motivated attack. Why? Because I have chosen to interpret everything in my world through the filter of race, and every event that occurs will in my mind be race related.

Note this applies just as well to any filter, christian extremists will interpret all events that happen as a sign of the times and the devil coming to power. If my filter is the class struggle than I will see every issue be between the haves and the have-nots.

My point is not to discredit people that want to bring gender and race in to everything I think there are many discussions where examining the role these factors played is important. My point is to caution and warn against getting into a single filter mindset, that is the cause of much of the divisiveness we see, we have people that really are living in totally different worlds because the way they choose to view the world is totally different and they refuse to ever consider there might be other filters to view the world through.

Or let's cut to the chase, you saw the headline, and your immediate conclusion was this was a raging SJW. Why did you jump to that conclusion?

https://science.sciencemag.org/content/360/6396/1465

These days I see a lot of both. I also have no idea what happened in this specific case, and at this time I wouldn't be willing to take a side.

You say "people with those types of built in personal issues float to the top" aren't you implying their positions are not earned based on merit and, therefore, that they are less competent than their peers?

After 2012 we got a diversity hire on our team who was useless in every way, was a pathological liar and manipulator, yet who was consistently rewarded because the bosses boss was a female feminist. Her boss knew she was trouble but could do nothing. I did encounter a few more cases where female employees would just inexplicably have skills gaps that shouldn't have happened at Google, like not knowing what hexadecimal was.

I never encountered grumbling about competition. Google was not a hugely competitive place at that time because there was more than enough for everyone. Promos were generally not quotad, for instance, at least not for most of the rank and file.

I was a high performance underrepresented minority, there's no way I'd go to Google. Wouldn't people just assume I'm there because of quota filling, instead of my actual performance?

If you were genuinely high performance then no, of course people wouldn't assume that. And sure you'd go to Google, because they pay very well.

The real issue and it's not at all Google specific are the minority low performers or outright troublemakers, like Gebru. Then people absolutely assume they are quota filling. They go anyway because they aren't self aware enough to realise that's happening and blame any lack of respect they get to racism/sexism. And the money is still great so why wouldn't they go?

For what it's worth, I and many others disagree with this characterization. I find Yann comes across as incredibly condescending and holier than thou in their interactions. Nor does Yann engage on substance. He refuses to take the time engage with, or even acknowledge that he has read, the relevant academic literature (which Timnit repeatedly cites).

At the same time, the narrative that someone above her in the management hierarchy decided to pull her research seems more plausible. In that case, I would tend to "Privacy for the Weak, Transparency for the Powerful." What you describe is the former, but this would seem to be a case of the latter.

>That's how it is supposed to work. The thing that sucked about it was that it was impossible to defend myself. I only found out details after the fact because the complaint was anonymously filed on behalf of a co-worker and it was the co-worker that later came to me to explain.

[1] https://provost.uchicago.edu/sites/default/files/documents/r...

How, exactly, is it supposed to work? If you are asked to respond to allegations, yet those allegations are not shared with you, what are they actually asking you to do?

Are you sure? Even primary school bullying is better, at least you can see the abuser.

>> Each rejection came with a new set of reasons. After some point, it started to seem to us that maybe academics just found this topic and our results too threatening. Maybe this paper simply was not politically correct.

It looks like the only reason to believe the reasons for rejecting the paper were political is that it was rejected by many venues. Note that having different reviewers give different feedback is par for the course.

Personally, I don't think this is serious. Having one's paper rejected, stings, but it happens to everyone and all the time. There's no reason to see hidden motives behind it. The paper not being very good is sufficient explanation.

Not that it doesn't happen for papers to be rejected because they threaten a reviewer's work. But to claim that the entire academic establishment is threatened by a paper and that's why it was rejected? Sounds a bit far-fetched.

Actually, I really can't think of how my advisor would tell me not to publish a paper, except to suggest that it's not ready for publication.

But perhaps, as is often the case in these conversations, we're talking about different "academia"'s? I think there are differences between fields and between different countries' institutions in the same field etc.

Mostly, I see in the article an attempt to engage in the usual online culture war, but I can't say what is reported in the article is making me concerned about academic freedom. To be honest.

People invested in identity politics are, by their own admission, interested in outcomes. In this world view, the legal system is a tool for achieving the desired outcomes. The same set of rules can be illegal if it rejects certain identities in a context, and at the same time legal if it promotes those identities in a different context. 'Heads I win, tails you lose', but dead serious.

A thread discussing the pervasive 'women and minorities encouraged' mindset is in western academia:

https://academia.stackexchange.com/questions/95011/is-it-con...

I have however seen way to many doctorates that seem technical while they're not. There's a common theme in her career, and it's not necessarily technical knowledge.

She worked at Microsoft research, sounds really impressive. But it was in the "Fairness, Accountability, Transparency and Ethics in AI lab". which sounds like a PR stunt. Her research was supported by Stanford DARE fellowship, which is mainly concerned with increasing diversity. She was an AI researcher at Google, but it consisted in pointing out bias in datasets. Her PhD research[1] was using street view images to find pick up trucks and finding a correlation with republican voters, which sounds more like a sociological application of computer vision than a hard technical problem.

[1]https://www.pnas.org/content/114/50/13108

I disagree because I don't think Big N has exclusive access to the data necessary for cutting edge research. You don't need petabytes of data to train cutting edge neural networks.

It also seems she published the paper she is best known for before joining Google, suggesting academics are not submitting subpar research compared to Big N.

Only highlighting issues without power to enforce anything is not enough to have any tangible results? That's for sure.

But she was not the CEO.

In general, if I hear "withering e-mail" without context, I imagine an e-mail sent between two parties (or at most, a handful) where one party is cutting down the other directly. When I hear that an employee was fired for a "withering e-mail", I immediately think of retaliation.

In this case, "e-mail" is almost inconsequential. This "e-mail" was a broadcast to hundreds of people who were not in any way involved with the subject matter. It was "withering" in the sense that it cut someone down, but the person/company it cut down wasn't even in the TO field.

Words matter. I now notice that the post title is changed to the much more neutral "AI researcher Timnit Gebru resigns from Google".

It isn't really clear with your first quote if one thinks that silencing people generally is okay, but not minorities/women, or if silencing anyone is never okay, or if the minority/woman was silenced merely for their demographic status.

Other companies have tried. Names and vitals deleted from resumes before any people involved in the hiring process see them, remote interviews conducted over IM or with voice masking. That's one benefit I see with the much-maligned leetcode-style interviews, is that your identity is completely isolated from the process until the very end.

This is actually kind of funny to me, because in the open source world plenty of people have extremely productive working relationships without ever seeing or hearing each other, and in many cases without even knowing what country the other person lives in, how old they are, or what their legal name is.

So while there are definitely arguments that face to face communication is "higher bandwidth" or has other advantages, it doesn't seem out of the question to me that the hiring process could be "blinded" to a similar extent to orchestra auditions, without any significant reduction in hiring accuracy.

(Ok, maybe not quite the same extent; language fluency and style of speech are still significant signals even if everything is done over text)

There's no reason you can't do an interview over text-chat instead of in person -- hell, with most of us working remote for the foreseeable future, that's how the majority of on-the-job communication will happen anyway.

Be careful here, the wind has changed and blind auditions are now problematic.

https://www.nytimes.com/2020/07/16/arts/music/blind-audition...

How is this true? What if someone is a broke straight white man with three kids and someone else LGBTQ person of color is a young childless heir of fortune? How is the first person possibly privileged in comparison with the second one?

Looking from Europe, America seems to be very determined to recast its very obvious class problem as a race (or even gender) problem, even if it means introducing disadvantages based on immutable characteristics AGAIN. Two wrongs do not make a right and this farce will bite you again in the future, just like slavery and old style racism did.

How fortunate for you. On the other hand, other people are hurt by the explicitly racist policies you endorse. You don’t speak for them.

> in my own caste

Why don’t you speak for yourself, instead of speaking for others and assuming they’re “in your caste”, whatever that means?

Imagine being a senior researcher at the peak of your career, working for 6 months to a year on a paper, submitting it to a top tier conference, getting accepted, then getting the pointy haired boss do-what-I-said-peasant treatment.

Big nope.

And I've lost the link to the tweet.

I've been on Hacker News over 11 years. I would like to think I've been building bridges, but it seems like nothing I do is ever enough to reassure people I'm not some SJW Feminazi here to just piss on the guys and let them know what misogynistic assholes they are and it gets hard to keep trying when most of the community has watched me starve for years and openly told me "Not my problem. Get a real job." (while my writing hits the front page, but people don't want to support it financially, knowing I'm handicapped, etc).

I don't have a whole lot more to give and I've spent much of this year wondering if it's time to throw in the towel and leave HN. It feels downright abusive at times to stay in a community where I feel I have done so much to reduce sexism and open doors for other women and improve participation of women here and it's the funnel for a multi-billion dollar business, yet I remain dirt poor.

It's a rather jagged, bitter pill to swallow and I think I deserve a helluva lot better.

But then I don't know where I would go if I left. HN is the least worst thing. Other places are worse.

Metafilter was a toxic cesspit that banned me for supposedly "self promoting." They like to wrench their shoulder out of place patting themselves on the back for how what awesome, wonderful people they are and the mods were actively encouraging the membership to bully me at a time when I was homeless. One member of Metafilter that had a hobby of harassing me while I was homeless was a female ER doctor. Another was a very privileged American pursuing their PhD while living in Europe.

When you left your comment, I was staring blankly at some other open tab wondering how in the heck to talk about rape prevention and best practices for dismantling rape culture without using the word "rape" at all, in part because it's a triggering word for people who have been assaulted and in part because I get accused of being full of bull for thinking I know anything about such topics.

Some problems are just hard to solve. I do what little I can, which doesn't seem to amount to a handful of sand in the grand scheme of things.

I guess the upside is I'm mostly well at this point, in spite of the entire world telling me I'm a deluded fruitcake so it's not like I'm ever going to get taken seriously or given any respect.

And this is probably all the wrong things to say, as usual. I wish I had an answer for you. There doesn't appear to be one, as best I can tell.

https://syncedreview.com/2020/06/30/yann-lecun-quits-twitter...

I really enjoyed this deck of hers, which I found as a result of reading this thread:

https://drive.google.com/file/d/1rcG8KVmjRUWWNSg-R6cTBlAScP9...

"So if you would like to change things, I suggest focusing on leadership accountability and thinking through what types of pressures can also be applied from the outside. For instance, I believe that the Congressional Black Caucus..."

Skipping parts of the chain of command to deal with grievances is usually a toxic behaviour. Trying to skip the entire corporation and going straight to the government counts too, IMO. This woman has opportunities, abilities and successes that most of us can only dream of and she's using that to try and make Google a vehicle for political goals not quite articulated here (this email is obviously not written for public consumption, it is light on specific details apart from her thinking there are insufficient women hired). I can see the complaint against her, she's already got an advantage in life and now she's using her access at Google to try and strengthen her hand and weaken that of her coworkers.

When I view Gebru's Twitter argument with Lecun throug the above lens, it is pretty obvious to me that she is behaving in a toxic manner.

In that fracas, she emphasizes emotional appeals and "you just need to listen to us" and "I'm so sick of this". She also takes a page from the AOC school of argument where you (the rhetorician) play the meta-game where you judge who is allowed to participate in discourse (hint: everyone who disagrees with me is excluded!). It's a useful trick, for once you have purified the field your arguments will easily win.

Lecun later offered an olive branch and apology, but, true to form, Gebru doesn't offer any hint of awareness that she also may be responsible for the toxic devolution of the conversation. She personally blamed Lecun for her getting trolled, whilst failing to acknowledge that he was also being trolled by her supporters.

Later, and separately, when Google calls her on her resignation bluff, she employs the "call out" tactic, where she accuses Jeff Dean of personally firing her in an obvious attempt to shame him. This whole affair reeks of a PR play (and thus, not acting in good faith). That, to me, is highly toxic.

Critical theory is useful for the purpose of study and quantification of different social bias, norms, tends, values, or whatever other social aspect you want to slice up. The problem is when people use those divisions to identify themselves. It forces people to identify with a certain group, when in reality the divisions are often not nearly as clear cut as they are defined. But over time people are pressured socially into moving closer to how their group is defined, and what may have once been a false dichotomy becomes a self fulfilling prophesy of sorts; where the social interactions are entirely defined by "my tribe" vs "their tribe". My tribe is all good, theirs is all bad, therefor we are justified in treating everybody in their tribe as less than human. Even is cases where there is very real injustice and evil taking place, allowing oneself to be defined by these groups does nothing to resolve the problem. It only serves to shore up divisions and conflict between groups, making it worse.

It is unfortunate that the admonition to "love your neighbour as yourself" is no longer held with high regard.

"Does that sound like a standard procedure to you or does it just happen to people like me who are constantly dehumanized"

She is not happy with the process and she is taking it personal. She was hired as a lead researcher and manages a big group and project. She has the nerve to say she is constantly dehumanized? 99% of Google employees would want to be in her position.

Google is a big company with many good but also bad processes and policies (I never worked there). Instead of trying to make change she starts wars. Her thing with Lecun was exactly the same. The funny thing she said she hired a feminist lawyer to sue Google a year before! She still kept her job after that. What company does that? She is lucky she had an employer like Google.

Did you see Jeff Dean's email they just added to the article? Not surprisingly, she is doing the same thing. The reviewers are not agreeing with her and she is shocked how someone dares to reject her paper. She starts making demands to reveal who that was and sends an email to the Women AI list like it's some racist attack against her because she is a woman and black researcher. That's a toxic behavior of a very arrogant researcher. She clearly expects to get some preferential treatment because she is a big shot researcher or she just thinks she is smarter than everyone else. The irony.

I mean, from the tweets what I gather is she wasn't planning on leaving just yet. Saying "hey, you want to resign by [X] but we'd rather you be gone so let us inform you've just been resigned as of now" is essentially firing [1].

> Offering to resign was a huge mistake.

I think this will be good for her in the long term. I'm sure she'll find another research position, and the alternative was betraying her convictions or just the huge burden of constantly being worn down by Google.

[1]: https://twitter.com/timnitGebru/status/1334364735446331392

Who determines what value a business gets assigned?

Industry research is not academia, where you'd expect to be free to choose the avenues of your research, and you should expect that publication is restricted. The technical improvements you make may be kept under wraps to gain a competitive advantage; and policy agenda work is essentally part of PR strategy, it's accepted if and only if that agenda benefits the strategic interests of the company. Tenure in academia offers some freedom of research (though not even then abolute), but when working for private companies, you don't have that freedom unless you have a specific contract asserting that, and even then you'd be wary of biting the hand that feeds you.

Like, if you want to expose climate change risks, it's not going to work out while employed in an oil company, and if you want to advocate for regulation of ML research while working in a company that would be bound by this regulation, this is certainly something that the management would be expected to care about.

Doing work that needs to be actively suppressed is not producing value - at least not to the company. So apparently there's a fundamental misalignment between the work Dr. Gebru wants to do and the work that Google wants her to do; this was made clear with the suppression of the paper, and escalated as Dr. Gebru apparently (and understandably) did not want to work in a very different direction, (quoting her tweet) "asked for simple conditions first" which were totally unacceptable to the employer ("We cannot agree to #1 and #2 as you are requesting."), so that's it, if neither is willing to compromise on their primary goals, they can't work together.

This does indicate that apparently listening to ethics researchers they hired is unacceptable to Google. Sad, but not surprising.

Context:

> Timnit co-authored a paper with four fellow Googlers as well as some external collaborators that needed to go through our review process (as is the case with all externally submitted papers). We’ve approved dozens of papers that Timnit and/or the other Googlers have authored and then published, but as you know, papers often require changes during the internal review process (or are even deemed unsuitable for submission).

And the key point:

> Unfortunately, this particular paper was only shared with a day’s notice before its deadline — we require two weeks for this sort of review — and then instead of awaiting reviewer feedback, it was approved for submission and submitted.

Not sure who approved for submission. Gebru herself or by someone in management? Then they address the weaknesses of the paper (as part of the review process).

> A cross functional team then reviewed the paper as part of our regular process and the authors were informed that it didn’t meet our bar for publication and were given feedback about why. It ignored too much relevant research — for example, it talked about the environmental impact of large models, but disregarded subsequent research showing much greater efficiencies. Similarly, it raised concerns about bias in language models, but didn’t take into account recent research to mitigate these issues. We acknowledge that the authors were extremely disappointed with the decision that Megan and I ultimately made, especially as they’d already submitted the paper.

Granted, its hard to track the timelines and who said what at what time, but there seems to be reasons why the paper would of failed their internal review process.

Also, let's be real. How much revenue are the vast majority of AI/ML researchers actually driving in these companies? Sure, someone gets promoted for multiple NeurIPS papers, and it looks great for the company. But most of the time, it doesn't directly impact products and revenue -- in the short-term -- and that's ok. Same deal for AI ethics and policy researchers.

I agree with this statement but for the life of me I can’t come up with a satisfactory answer as to why companies don’t want to actually change things. Is it purely institutional inertia? Are the benefits of a more diverse workplace not accepted by the powers that be, so efforts to achieve that are undermined?

"Look how awesome this activist is we're hiring," easily becomes "Look how unreasonable this employee is we're firing."

In this case, both Google (a social justice activist gets them into a public screaming match, what a surprise) and Ms. Gebru (played chicken with her employer and lost).

Even more than that, it's far easier in infosec to say "There is gatekeeping going on. I'm going off to make my own space where you're not invited."

Seems to be a very broad brush that you're painting with. "Progressive people" encompasses a wide variety of individuals and beliefs.

From what I can tell, she doesn't seem to address the points raised at all and instead complains they won't tell her exactly who said what.

How is this either/or? Highly educated people can still join fringe political clubs.

There is nothing about technical education that immunizes you against radical politics. At the risk of Godwinizing the reply: the engineers who built the first liquid fuelled rockets in Peenemunde were N..z..s, including the guy who was later brought into the US and built all kind of powerful and ingenious rockets there (Wernher von Braun).