Trying to get context, I see big names professors from Stanford, Caltech, Berkeley, Cornell all commenting/retweeting on how bad this is, so it's a big deal for sure
It's pretty common in large corporations to have 'affinity groups', which discuss topics that affect the group.
I don't work at Google; I work at Amazon. I can only speak about my experience, but it's likely the same at Google.
You don't have to be a part of an affinity group to be a member of the mailing list (that's where the ..and Allies portion comes into play).
The 'Google Brain Women and Allies' mailing list is probably full of women from the Google Brain team, non-binary and men from in from the Google Brain team, and others who aren't on the Google Brain team but are interested in its work.
Again, I don't know. But this is educated conjecture based on my experience at another FAANG.
Must be a US specific thing as I do not see this in EU corporations. I assume there is also a 'Google Brain Men and Allies'? Why not just a 'Google Brain and Allies'? Or does that also exist?
I would assume that there isn't a 'Google Brain Men and Allies' affinity group. "Google Brain" does likely have its own general distro where standard work-related topics are discussed.
At the hazard of becoming 'political' and triggering people, affinity groups are typically created for groups that have traditionally been underserved, marginalized, or have special considerations.
As a cis male, I don't really see the need for a 'Google Brain Men and Allies' distro list because the issues that men face in the workplace are generally the same issues everyone faces. Women (and nonbinary people) can be confronted with issues that men simply aren't.
---
At Amazon, there are groups for Black employees, Latinx employees, LGBTQA+ employees, transgender employees, women, disabled employees, etc.
Sometimes these groups talk about work; but sometimes they talk about just life things. For example, often cisgendered employees will have children that come out as trans. So they'll reach out to the transgender community for help and support as to how to be the best advocate for their child.
Or, as a personal example, I moved to a new city to take the job at Amazon. I was looking for a therapist, and it was important that the therapist be supportive of LGBTQA+ issues and ethical non-monogamy. So I reached out to the LGBTQA+ group for therapist recommendations.
I understand these seem to be sensitive and 'political' issues in the US. I just do not understand how any of this labeling and segregation can be productive in a workplace.
You don’t understand how trans people might have specific things in common to talk about that non-trans people don’t relate to? These groups aren’t just for political hot topics, but also for parents or for musicians or runners or any other affinity group.
I can't speak for Google Brain specifically, or AI/ML (I do InfoSec). Perhaps there is a need for discussing the Google Brain project specifically from a woman's point of view.
But when I hear about a 'Google Brain Women and Allies' distro, I'm more thinking about peers which discuss things which impact them. I wouldn't get too hung up about the 'Google Brain' portion of the distro; it's likely just a way to self-select a group of peers which share a common interest.
Ahh I got it now. This is great answer and would make a lot of sense. The tech industry is very male-driven and plagued with a lot of issues, so I’d only imagine these groups as essential part of many employees survival / wellbeing.
Also I’d argue that these are people issues and not US political issues.
The list is presumably a place to discuss issues that are specific to women’s experience working on Google Brain. “And allies” indicates it’s not exclusive, and other folks are welcome to contribute to the discussion. The reason there’s no “Google Brain Men and Allies” is because “men’s experience” is the default around which every other mailing list likely focuses.
Why would workplace mailing list focus by default on "men's experience" or "men's issues"? Wouldn't they just focus on work stuff, irrespective of the gender of the persons working there?
In the US, men aren't considered marginalized, and therefore having Male, White, or Straight affinity groups is frowned upon. They are often seen as opposition to Gender/Racial justice movements. People who bring up these issues are often labeled fragile or racist.
In my opinion, this leads to division and contempt/hatred for the other group. I live in a super progressive US state, and the university near me got a lot of backlash for hosting White affinity groups.
All this says is, she sent a mail to Brain Women and Allies that resulted in her immediate termination.
Does anyone else remember when she picked a fight with Yann LeCunn by deliberately misinterpreting his remarks to cast him as a racist?
Seems reasonable to assume she picked another fight. Reading the Twitter threat, it reads like she made demands in a way visible to a large number of employees (something about authorship on a paper, to skim the Twitter thread) and perhaps attacked Google while being a manager there (the emphasis on talking to non-management employees sounds like this). I assume the email will leak, this being Google we are talking about.
Says she deliberately attempts to cast LeCunn as a racist is a very uncharitable framing, but putting that aside, it doesn’t actually seem reasonable to assume she picked a fight. We have no idea what was in the email.
How would you describe what happened? LeCunn calmly and rationally explained that a particular model produced biased results, and that that this was due to the bias in its training data. He was then excoriated as a racist by this woman and the Twitter mobs she incited, to the point where this Turing Award winner is no longer willing to engage with the public via that medium.
LeCunn made repeated attempts to explain his position, only to be virtually shouted over about he wasn’t listening to people of color because he pointed out that a model trained on White people learned to produce images of White people, and that if it had been trained on Black people it would produce images of Black people. It was a research paper, not a production system, and Gebru and her acolytes were much more interested in scoring cheap points than in having a serious conversation about either ML fairness or the merits of this particular paper.
Seriously, the primary sources can be read by anyone.
> LeCunn calmly and rationally explained that a particular model produced biased results, and that that this was due to the bias in its training data.
But this claim is certainly not necessarily correct, and he shouldn't have been so confident about it. Any part of a system can contribute to bias and that includes the model design, not just the data. If they want the system to work then they need to actually demonstrate it works, not just say it could with X change without testing that.
Though I remember people arguing it didn't work properly at the same time they were saying it was evil (because it contributed to surveillance). Which is odd because if you don't want it to exist, you shouldn't want it to work either.
Wouldn't that go both ways though? Why was Gebru so confident that the bias is because of something else other than the training dataset? She immediately assumed that there are other items contributing to the bias and devolved the quality of the conversation as well by making those claims in a frustrated/exasperated tone.
The most obvious and easiest way to clear this up would be for the original researches to train the model on a "Senegal" dataset as mentioned by LeCunn and see what the results are.
The whole point is that it never is trained on a "Senegal" dataset, is it? Why do we always see these errors where white-centric datasets produce white-centric results?
While it would be a symmetric situation a vacuum, we do not live in a vacuum. And acknowledging dataset bias by itself doesn't address the bias meaningfully. In practice, we often treat the bias as an exogenous factor when it is not, moving it outside the scope of our responsibility. But it is very much the product of our work, a reflection of our choices, values, and beliefs about what to prioritize. We can't abdicate our responsibility for it; the stakes are too high.
Well that is the whole point. The entire controversy could have been cleared up if the original researchers or even Gebru would have trained the model on a "Senegal" dataset, or a dataset diverse enough to their own liking and then make the findings public. That, to me is research. Throwing a fit on twitter because people won't buy that the outcome of this model is racist is not research.
I am not sure why the wording around this has to be so abstract.
This is a great goal but the time and place for it is not in a super-resolution paper. There are specific conferences, benchmarks and projects where it can be achieved with more efficiency.
No. The easiest way to clear this up would be for Gebru to make a specific claim that was actually related to the model and include a pull request of her own modifications that she had tested to deliver superior results.
This would be called “research”. That is what she was hired to do at Google. For everybody of every race ethnicity orientation able status and background that would be honored to give their best effort at the opportunity to do research at google, acceptance of resignation was the right thing.
Of course. Or at the very least, she herself could have trained the model on a "Senegal" dataset and make the findings public.
My only point around her feud with LeCunn is that she threw a fit because people including LeCunn pointed out that it is not necessarily a race thing, and she wanted to give it a race spin without confirming that it was actually the case.
And I am not at all commenting on all this drama around her resignation/being fired without getting a look at the contents of the two emails: the one she sent to Brain Women and the one with her demands. But I know this: if I were to issue an ultimatum to my employer, I am treading dangerous waters and should be able to digest the outcome of that, including being fired.
I think it is reasonable for one of the founders and most highly respected members of the field to make confident claims about something that is as open and shut as this. The paper in question was training a network to upres faces. The researchers trained it on a dataset containing primarily White faces and released it on GitHub for reproducibility. Subsequent people used it on Black faces and found it make them white.
This is not surprising. Black faces and White faces are not the same data manifold. This is like training a network to upsample oranges, then running it on an apple and being surprised when the result is an unusual color and texture for an apple.
I’m not sure what else there is possibly wrong here. Is the width of their convolutions racist? Their choice to work on super resolution? The fact that they released their work for reproducibility?
> I’m not sure what else there is possibly wrong here. Is the width of their convolutions racist?
Not that parameter specifically, but if you say the data is wrong then that means any other parameter might be wrong too. More data might mean the model is too small to fit it, or the hyperparameters might be wrong to train it, or you now have the wrong ratio of other phenotypes (let's say that instead of races…) in the training set and their results regress.
Also, if you're adding people with darker skin, that of course means the pixel values are lower. That matters for image processing code, things like SAD thresholds or noise reduction will work differently.
> Their choice to work on super resolution?
Superresolution is only useful as a toy and they should have mentioned that when they put up a live colab, yes. They added a disclaimer later on and it was good - there was a big issue where people were convinced this was somehow a surveillance technology because they watched too many TV shows, even though it literally can't work that way!
The data isn’t a separate thing from the rest of the model (unless you’ve validated it works across different datasets). It’s all just a bunch of numbers that get jointly optimized.
>This is not surprising. Black faces and White faces are not the same data manifold.
in particular because features contrast (dynamic range) is lower for darker faces.
>Is the width of their convolutions racist?
Not width. As a result of the above mentioned lower contrast, the racist here is the sensitivity of the resulting Gabor filters produced by the training in the first convolution layers and the Gauss filters in the next layer's. I suspect to deal with that problem one would have to normalize dynamic range of the faces, i.e. it would look something like kind of lightening of the dark faces and/or darkening of the white ones.
There is a discussion to be had here but it's not possible in public because of the escalation of the topic. This is also not the best thread to be having it.
Since everyone else is just saying "you don't get it" without explaining what "it" is, I will provide some brief avenues of exploration. Because they are brief, they will be coarse and imperfect, and aiming to give you directional assistance on the topic.
So with that massive preamble because I'm not seeking to argue, here you go:
* Pre-trained models already encode much of this bias and if you don't use them you won't get very far very fast
* Large available datasets also reflect these biases
* AI model performance is generally measured against this dataset, further entrenching the bias. If you're better on Indian face generation it will give you barely any benefit on most scoring methods
* Saying "the outcome is only biased because the data is biased" misses the fact that the performance is on the biased data
There's a little more to it but I believe that's the meat of it. Anyway, not too keen on arguing this. Just sharing because it took me some work to figure out what they were talking about and I wish someone had explained it to me so doing so here.
Thanks for the additional detail, but those all still boil down to dataset eventually, right? Those sound more like "yes, and" additions than the "no, you're racist" shitfight we saw.
In the interest of explaining (not arguing), I believe the distinction is sort of that the community at large recognizes performance only on certain metrics. Let's set aside the notion of whether something or someone is "racist" because that's a bit charged. We'll instead just talk about whether we're creating generally capable systems.
For instance, if I were to champion metric A which purports to measure performance on human faces but it really only rewards performance on Senegalese then models that do better in general may not be recognized for being better.
In an isolated sense this is not a problem. However if the mainstream is that all metrics that are taken seriously are dataset-biased then we'll have an environment where the models will be trained on biased datasets in order to be successful.
For instance if everyone uses LFW to determine how good facial recognition is, then Senegalese fine-tuned facial recognition tech will not be recognized as being good at facial recognition.
So the argument is that dataset bias is built-in to our approach to the problem. I, personally, think that this isn't malice. We need benchmarks to judge approaches against each other. Benchmarks always have a first mover advantage and a massive path dependence issue. The first benchmarks aiming for things on humans do not reflect humanity accurately. These benchmarks became standard among the community. To be taken seriously you have to do well on benchmarks that are standard in the community. Dataset bias is then natural in newer approaches because the approaches are judged against how good they are against the inaccurate (if you will) benchmarks.
So no one need be racist or anything for the end result of the field to end up being discriminatory.
I don't think a successful approach is to call people racist over this. After all, it isn't malice that guides them. The discrimination comes from the sort of historical accident that has North facing up on a map. And no individual is really racist. It's sort of like the Bechdel test: no movie is crappy simply because it doesn't have two girls talking to each other, but if very few movies have two girls talking to each other about something other than boys, then it makes you think "hmmm, why's that the case".
Wouldn't it be a lot more productive to lobby for inclusion of more samples into the performance measurement datasets than to get into super-combative twitter pissfights with people who aren't even opposed to that?
That's, like, not even culture war, it's just basic correctness of the reference datasets. If a fruit classifier was missing oranges, we'd just fix it and move on.
Haha, well, you know how it is. Everyone has their own opinion as to how to solve problems like this. Some people see their role as sort of fire alarms to say "Hey! This thing is happening!". You know, like how every thing has people at all stages in the process. Others might believe that this is one instance (the benchmarks being 'wrong') and that a better outcome would be that the people putting out this stuff are aware of ways that things can go wrong so they think about it in future (maybe this time it's these benchmarks, next time it's the other ones, and then maybe next time it's something inherent to the structure, whatever, it's a bit whack-a-mole to solve at the instance level).
Like, for instance, HN has people who will bring up privacy violations of big tech constantly. They see their role as making sure the conversation is happening. Not justifying this. Just aiming to understand it.
For my part, I prefer to take the approach you're talking about because I, too, think that the fastest path to this is getting the photos, labeling the photos, and then lobbying for inclusion. Ultimately, I think it's okay if things optimize fast for growth and then we fix up issues afterwards. So the people building the benchmark sets weren't able to get a set that's representative of humanity. Should they have waited till they could have done that? IMHO, no. Rapid release moves the state of the art forward and then we can put in all of these corrections as we move.
Then there's the question of whether all-humans dataset is a good thing or if instead a thing that is white-humans and another that is black-humans is better. Anyway, all said, my personal approach to this problem (if I cared about it a lot, which I don't) would be to say "Current benchmarks and training data available bias towards certain races. I'd like to build X/Y to solve that. Here's what I have so far" etc. etc. I think positive engagement like that yields better results because the vast majority of scientists actually aren't weird race supremacists and the vast majority of AI researchers will gobble up any more data you give them which is segmented and labeled differently, the greedy bastards :D
Some challenges that I can still see:
* Getting the data. Might not actually exist.
* Labeling the data. Probably needs some work.
* Getting it into the benchmarks. It'll invalidate old scores, so there's just a product adoption problem here. I don't know how the community handles newer releases.
As a last aside, I suspect that this conversation ended up the way it did because:
a. It's charged. It's race-based differing outcomes. That's a sensitive subject.
b. People feel unheard. This is natural. Like, this is not an 'interesting' problem. It's literally just a data error so the luminaries in the techniques part of the field aren't really that interested in it. And the techniques part is where the sexy is.
c. This sort of thing has a tendency to escalate. One side says "You're not listening to what I say" and the other side says "I'm not racist. I don't get why you're calling me that" and before you know it it becomes "You have to be racist to be ignoring me" and whatnot and de-escalation becomes impossible. Especially because everyone rewards the loudest on each side.
Honestly, I think it's quite interesting to observe and to understand as just a view into the human condition but we use AI models professionally in the GIS space and professionally we just don't go near this at all. No part of me finds it interesting to solve or to interact with the discussion in any way. I only sort of participated in this here because I think I managed some insight into what it is and I wanted to write that down because I wish someone else could have accelerated me into it.
Anyway, I think that's all the insight I have on ...
I'd never heard of this fight before, so I just looked it up. My interpretation is that he made that claim, Gebru responded by claiming that that was an overly reductionist view of things (given, presumably, the fact that biased assumptions could affect the coded assumptions as well as the training data), he maintained his opinion, and she felt frustrated that she didn't feel she was being listened to. I did not see her call him a racist, and I did not think either person was acting in bad faith; both seemed frustrated and unhappy that there was a disagreement.
I think it’s clear Gebru was acting in bad faith and LeCunn was baffled and trying to deescalate. Gebru responds to each attempt to deescalate by a reply designed to further rile up Twitter, without actually talking to any of the points LeCunn makes. Virtually all of her replies are some variant of “you are wrong, but I won’t say why” and “listen to me because I am Black”.
But that drama was some time ago. She should have been fired on the spot back then, but, at Google women are virtually unfireable, black women doubly so. Whatever she did to get canned like this must have been extreme. Either that or Google is learning how to hold these sorts of employees to the same standards as their less fashionable colleagues.
Edit: incorrectly claimed the spat was before 2020, it wasn't. 2020 just feels like an incredibly long year.
What do you expect me to cite to show that? I worked there and saw it for myself. That's the citation. But really, do you doubt it? It's a company that talks constantly about how hard it tries to boost the number of women and blacks working there. Firing them works against that goal.
On a scale of 1 - 10, how angry do you get at the thought of a black person arguing with you? You can be honest. How hard does it get your forehead vein throbbing? What if she's also a woman? I'm betting the very thought makes you turn tomato red.
Ah yes, the classic "if you're against affirmative action you're a racist" line. Knock it off. I'm the opposite of racist: the thought of arguing with a black woman doesn't bother me in the slightest. I'll happily argue with anyone.
The racism here is in the behaviour of treating some people specially because they are of "wanted" races or genders. I'm criticising racism, not engaging in it. But clearly a lot of people especially in the US are so far gone they'll never recognise it.
It seems like most of the other AI researchers, at least those mentioned in that summary, agreed with Gebru? So while Gebru may have been “acting in bad faith,” it seems that the consensus, at least as the article presents it, was that Yann was wrong.
Maybe you’ve noticed that almost everyone who thinks this was probably justified is using a throwaway, and everyone who is claiming it was unjustified is not. Only one group stands to lose their careers here.
The discussion shows that Gebru is a racist to the bone because she viewed everything from the racism angle. LeCunn could be wrong by attributing bias to statistical bias in data, but how is it bigotry without evidence? Yet, oh no, LeCunn is a racist in Gebru's eyes, and in this strange American culture, Gebru could just call anyone or any opinion she didn't like racist or racism.
I'd say she deserved to be fired for she is a racist who's damning to a normal society.
Racism is discrimination based on race (from a position of power)
Whom did Gebru discriminate against, based on their race? No one.
That episode is unfortunate imho, but she only suspected racist rationales... Calling out racism does not make you a racist, just like calling something "fishy" does not make you grow fins and gills
Calling out racism does not make one a racist, for sure. Labeling anyone who disagrees with her a racist, without even a slightly plausible evidence, makes her a racist because all she can see is race. You're right that that's not the standard definition of racism. It's a fair game, though, as people like Gebru pretty much uses racism as a weapon however they like.
Fair point. "Any" is too strong a word that invites just one counterexample to get debunked. I'd say "in many cases" then, as demonstrated in her Twitter stream and her email. The words like "white men"( such a vile and irrational perspective), "oppressors", "racists" show up often. Case in point, a professor who tells his/her student that the student's life was easy, even though the student had to sleep only four hours a day. Lack of empathy, sure. It's racism without other evidence? Are you kidding me?
And her paper was rejected and her immediate reaction is to demand the company to reveal the identity of every reviewer? Yeah, right.
I have to disagree. I read the article posted above and this strikes me as an excessively critical view. I don't see Dr. Gebru acting in bad faith although they are clearly frustrated with the exchange.
> Virtually all of her replies are some variant of “you are wrong, but I won’t say why” and “listen to me because I am Black”.
In Gebru's final thread on the topic (https://twitter.com/timnitGebru/status/1276326634195259392), she lectured LeCun how to give a proper apology, stated that he was engaging in "pattern of marginalization" causing "incredible harm", and instructed him to educate himself and listen to Black and Brown people before speaking to her again.
That's a shame, she only needed to add "lived experience" and "mansplaining" for me to fill in my "woke bigot weaponising her race and gender to shut down a valid discussion" bingo sheet.
We've banned this account for using HN for ideological battle, which is against the site guidelines and not what this site is for. We've also banned a bunch of related accounts that were doing the same thing, including posting in the same thread. That's seriously abusive.
Please don't create accounts to break HN's guidelines with!
She has been advancing her agenda for a while by strategically picking fights, and winning them because she's been on the right side of a political wave and knew how to ride it. Then she got blinded by her own success and overplayed her hand, a classic cautionary tale. Fun fact: many people, not necessarily managers, won't be shedding tears seeing her go.
It's not "reasonable" to assume "she picked a fight". It's not "reasonable" to fire someone for "picking a fight", whatever that is. It's awful how people are willing to be so charitable to Google here after doing the exact opposite in Damore's case.
It's not really about charity to Google. I just don't consider her Twitter feed a trustworthy source of information, given her general frequency of toxic Twitter drama and her extended, bad faith attack on LeCun. I wouldn't be surprised to learn that Google was in the wrong here, and if they are I hope she gets some kind of justice, but I don't think we can just take her word for it.
People get fired for picking fights all the time. You say that's not reasonable, but what's reasonable about subjecting all the other employees at a company to somebody who picks fights?
At least Google was consistent: it fired both people who were causing too much trouble.
(BTW This is actually the fun part of the Damore story: the guy think he is smart by explaining how "evolution" makes gender different but misses the ABSOLUTE main point of evolution which is that "to survive to an environement you must fit in it". He didn't fit in the google environement, he got fired.)
The email she posted [1] makes it pretty clear that she sent an email with some demands/conditions (separate from the one "to Brain women and allies") and said that she would resign if they weren't met. Google said they wouldn't agree to them, accepted her threatened resignation, and told her it was effective immediately.
Without knowing more context, it sounds like she tried to do some hardball negotiating by threatening to leave, and got her bluff called.
Quoting the tweets for posterity, link at the bottom:
> I need to be very careful what I say so let me be clear. They can come after me. No one told me that I was fired. You know legal speak, given that we're seeing who we're dealing with. This is the exact email I received from Megan who reports to Jeff
> Who I can't imagine would do this without consulting and clearing with him of course. So this is what is written in the email:
> Thanks for making your conditions clear. We cannot agree to #1 and #2 as you are requesting. We respect your decision to leave Google as a result, and we are accepting your resignation.
> However, we believe the end of your employment should happen faster than your email reflects because certain aspects of the email you sent last night to non-management employees in the brain group reflect behavior that is inconsistent with the expectations of a Google manager.
> As a result, we are accepting your resignation immediately, effective today. We will send your final paycheck to your address in Workday. When you return from your vacation, PeopleOps will reach out to you to coordinate the return of Google devices and assets.
There is still a civil way to accept a resignation, with proper timelines and transfer of responsibilities. She offered to resign in some time due to some demands, and the immediate termination seems to be related to an email unrelated to those demands. This sounds a lot like a sleazy move by Google to couch an immediate firing as a resignation.
There's no legal requirement for a termination process in the US. A company may choose to give notice, but with a few rare exceptions, they can fire you on the spot at any time for any reason if they want to.
In many U.S. states to include California, employment is legally considered “at-will.” This means the employer can terminate the employment agreement at any time and for nearly any reason.
You cannot fire someone for membership in a protected class (race, religion, sexual orientation, etc) or participation in protected activities like reporting an employer’s violations of the law, but it’s difficult to prove intent in such cases.
There are exceptions for unionized workplaces with collective bargaining agreements. Employers in those situations cannot fire employees without “just cause.” Google R&D is not unionized, however.
Im most cases downvotes are knee-jerk reaction. I find it normal. People are sensitive creatures with strong biases as default. I read HN actively from 2008 (good old days:) ) nowadays I always look at downvoted comments first to gain information.
There is no alternative, and probably will not be another. HN is a result of optimistic hacker culture with big dreams for success and tons of useful info and people. Ideological and practical Impact of HN on my career (18 years +) is enormous, it is on pair with early Google, INFOQ, Nielsen Norman Group. We live in different times now, innovation is hard, most of the valuable users don't have free time to post and I understand this. You can still extract value from HN, but with more friction and effort.:)
Her tweet suggests she submitted a resignation letter, and offered to "take her name off of it" if they met her terms. Means that's not a threat, they got a signed letter for her resignation.
She made some pretty serious demands, not having been met, with such politicization there's no reasonable way to accommodate a longer handoff, and she knew that. Waaay too much risk there.
She set her own terms, they accepted, that's it.
She should have been more transparent in this, because in her Tweets she wasn't forthcoming that it was her terms.
There are a ton of people who are sympathetic to those who are oppressed, but misrepresentations, cancelling etc. on the wrong terms are going to lose general sympathy very quickly.
None of this should be construed as a measure of whether or not Google is 'Good or Bad' or whatever, all these things are individual situations, with their own unique circumstances.
Do we even know what her demands were? I haven’t seen them posted anywhere. That said, it’s not behooven on either party to be transparent — to the public, or even other employees — because that’s just lawsuit bait.
All we know is that she gave a hardball ultimatum, and Google accepted option 2. I don’t know enough about what’s going on, but if you say “Do this, or I quit,” you have to be prepared for them to show you the door immediately. That’s how at-will employment works.
> In the letter she criticised the use of pre trained language models in Google's products (e.g., BERT is now used for most searches, machine translation, etc.). Apparently, the two conditions she mentioned concern how Google goes forward in deploying these models despite the warning of (their own) AI ethics researchers about biases that are manifested in these models.
That seems like its not only reasonable, but what her job was. Now if she said, "Do this or I quit" then that is on her. But if she criticized them harshly then that seems completely appropriate.
There is bias in Googles original search - it crawled content. That content will be biased.
Google search has been biased since the start. It has nothing to do with AI.
Biased is also contextual: what does it mean for all of us non-Americans to see tons of American content in everything - the 'American bias' is overwhelmingly the strongest bias, where are the concerns about that?
And how does bias imply a 'lack of ethics'?
The entire consideration is ridiculous:
1) AI is not special and odes not deserve it's own ethical oversight. Every social tech has issues and it all needs to be thought about.
2) Someone with no real training in the issue may very well merely be injecting their own politics into the situation, and possibly overstating issuses.
3) There's nothing objective about morality or ethnics, so it's really hard to even find such at thing as 'objective'.
What the company needs is a clean, comprehensive framework for the issue, and probably some independent oversight that can given them private assessments of where there are red flags.
Not individuals who want to make a name for themselves on the issue publicly, and who might be part of some kind of ideological movement.
There is bias in everything -- that is why you need ethics. You need to determine which biases are problematic and which aren't. If Google search favored Chinese government results in the US you'd probably say, "that seems biased in a bad way".
Your disagreement seems less to do with that and more to do that you disagree with where her ethics rest. I think the reason why she pushes for marginalized groups is that those are the groups that are less likely to be able to stand for themselves and thus become the victims of the bias. A bias against corporate or powerful interests will likely be rectified relatively quickly.
Yes - the bias can be worse for marginalized groups and that's a legit issue for sure.
But hiring activists, and particularly those with a focus on a very specific technology ... I think is a double-wrong.
'Ethics' is all encompassing, every company has to deal with it ... it needs probably specialization and some cold calculating.
If there were some gigantic loophole whereupon marginalized people were truly thrown under the bus here, then maybe there's a case, but I seriously doubt that.
At least one of the demands was, quoting Jeff Dean's email, "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."
If someone says (in nicer words) "Do X or I'll resign" and the company doesn't agree to those terms, it's best to accept the resignation immediately because the employee's interests aren't aligned with the company's.
It's also pretty common in roles like sales and PR to be shown the door the day you give notice, especially if you're going to a competitor.
I'm surprised that she opted to publicly put the blame on someone and link to their Twitter. That could be interpreted as a dog whistle to her followers to start the public shaming.
But at the very least, I'd guess that this will make any potential future employer think twice if they really want to be her manager.
And lastly, I remember the name Jeff Dean from reading about advanced AI research, so if someone publicly says "Jeff Bean says XY is unbearable", I'd be inclined to just believe it.
All in all, this looks to me like Timnit Gebru could improve her career chances by being a bit more diplomatic with her public announcements.
Jeff Dean isn't just someone from advanced AI research. He's one of the longest time Google employees and a key part of the company's success. He's one of the chief engineers behind MapReduce, BigTable, Spanner, the re-awakening of neural network research and many other incredible achievements. At one point inside Google there was a website called "Jeff Dean Facts" similar to the Chuck Norris facts. The guys reputation is out of this world.
He's also a rather mild mannered and very nice man. When I worked there I never heard anyone say anything bad about Dean. He rarely expressed strong opinions even on engineering topics; the sort of guy who prefers to win debates by simply coding up the best solution.
The name Timnit Gebru also rings bells for me. Last time this name came up she was busy chasing Yann LeCun off Twitter with pitchforks. LeCun! One of the other major luminaries of AI research. In fact she stirred up so much trouble and made so many outrageous accusations he quit Twitter altogether. She appears to be a serious troublemaker, far more interested in identity politics than AI. She was an embarrassment to Google before and it's good for them they finally canned her, but seriously, why did it take this long? And who will hire her next? I'm sure someone will be dumb enough to do that but hardly any companies have openings for "ethical AI researchers", and she already burned her bridges at Facebook.
I'm not sure why this is downvoted? I fact checked the Yann Lecun claim and I learned something new. Similar to Stallmans pitchforking, it's a sad case.
I second the fact that Jeff is an extremely mild-mannered and kind (and supportive to young employees) person. If he reacted this way I assume his patience must have been worked to the ground...
She is being dishonest from what I can tell. She thought she could just boss people around without consequence but now she is - hopefully - finding out that isn't the case. I really hope Google doesn't buckle under Twitter pressure...
>There is still a civil way to accept a resignation, with proper timelines and transfer of responsibilities.
that just reminded how Marisa resigned - if i remember correctly she just called on Monday morning to notify that she is already at Yahoo. So the "at-will" works both way.
>> certain aspects of the email you sent last night to non-management employees in the brain group reflect behavior that is inconsistent with the expectations of a Google manager.
as my friends managers explained to me - a manager represents the company to employees, i.e. when a manager speaks it is basically the company speaks and thus [until the company reacts quickly by disowning the manager's words] the company bears the burdens and responsibilities that the manager made promises/representations about - similar to that relationship between Pope and God pictured in the Dogma - so managers are trained to be very accurate with their words, and understandably a manager going off-rails is like a fire emergency to be dealt with immediately.
You're right but, the tone of the mail hints that there's an underlying tension already. Also, the mention of other email (which I didn't read) further hints that there's something heated going on amongst the parties.
So looks like Timnit sent an e-mail to the group which wasn't OK in the eyes of the higher management, then sent a strong mail about her demands to the same management and used "I'll resign otherwise" card to force her way.
The management, agitated from the first e-mail, didn't buy it and sent her home. The way they did is neither elegant nor polite but, she pushed a lot to get this response it seems.
This is my understanding of the issue and I may be completely wrong but, this is what I was able to decipher.
I don't think she ever resigned. My guess is that her manager Megan was being nice to her by saying that this is not firing but "accepting resignation". Unfortunately, her manager was obviously not trained in firing. Never "accept resignation" if employee didn't actually resigned, even if you want to nice to them.
If you say “I will resign unless X”, your employer is then entitled to accept your resignation. HR prefers to accept resignations rather than fire people, because then the company is much less likely to be found at fault, and it sounds like this communication was carefully managed with the help of HR.
> Without knowing more context, it sounds like she tried to do some hardball negotiating by threatening to leave, and got her bluff called.
Or she was on her way out anyway (or at least ambivalent about staying, it sounds like) and wanted to provoke Google to fire her. These days getting fired as a female minority is a good way to elevate your reputation among a certain crowd, maybe get a more lucrative offer elsewhere, and possibly get a payday if your employer didn’t have good lawyers.
Just look at her Twitter mentions right now, she’s getting an enormous amount of attention, including job offers and lawyer recommendations.
>These days getting fired as a female minority is a good way to elevate your reputation among a certain crowd, maybe get a more lucrative offer elsewhere, and possibly get a payday if your employer didn’t have good lawyers.
As I understand, she first sent out mass email to large group of employees that outright violated Google's policy. This act alone was sufficient for immediate termination. However, she then tried to bargain with her manager. Her manager basically let her know that bargaining was irrelevant and her action was serious enough that she is being terminated immediately. Then she comes back and tells the story with twist that Google ignored her bargaining and terminated her even while she initiated that bargaining. She goes on to reveal a lot of things except the content of her email or even hint of exactly what was so incriminating. Now she wants to keep lid closed in the name of future lawsuit all the while launching attacks on everyone. I feel whole thing is super muddy. If she is truly ethical, she needs to come out clean on exactly what happened or just stop attacking everyone and keep playing victim.
The email has been posted in full and it sounds like it was to a standard mailing list. Nothing in the email sounds like something you would be fired over. But Google seems really sensitive these days.
after 20 years at several US BigCo-s i see such an email as a totally fireable offense for a manager in a US company (and in Russia the boss would do bad things to you and would still probably fire you after that if you don't come to your senses :):
It also sounds like she did lawyer-rammed the Google a year before, so they were tip-topping around her strictly by the book, and with such clear an offense as that email she just threw the HR a freebie, as another commenter put it, so they could now get rid of her strictly by the book.
Could you outline what content in the email as an "outright violation of Google's policy"? Having read the email, it did not strike me as a fire-able offense.
> Have you ever heard of someone getting “feedback” on a paper through a privileged and confidential document to HR? Does that sound like a standard procedure to you or does it just happen to people like me who are constantly dehumanized?
Taking a dispute with your management chain re: feedback that said management chain felt they had to take actions like a private, hr-involved feedback session, and broadcasting that dispute to a mailing list of peers, downlevels, etc.
> But now there’s an additional layer saying any privileged person can decide that they don’t want your paper out with zero conversation.
This was, from the perspective of google management, probably very much not a reasonable characterization of the aforementioned circumstances. Managers represent the company, including publicly. Timnit was obviously in an awkward position where she seems to have been a player coach, ie a manager also doing IC level work (in this case, research). It feels like she expected her actions as an IC not to be viewed as actions that also came from a manager. That's a very hard dual role to have.
Third, maybe shouldn't get you fired in a perfect world, but prolly will in ours:
Public internal criticism of DEI efforts on the part of your management chain. Imagine how that reads to a line-level employee at Google. Particularly an URM.
Fourth, same vein -- auguring on an internal mailing list
to get the CBC to criticize your employer
> I believe that the Congressional Black Caucus is the entity that started forcing tech companies to report their diversity numbers
I'm only paying attention because I'm curious if Timnit was hired to -- if you will excuse the phrase -- whitewash google's policies, and she didn't understand what Google was buying; or if there genuinely was some conflict here. I suspect both from different players in the executive ranks.
Finally, someone who got what this is about; attacking anyone and everyone that dared to question her right to attack the company that employed her and the people that work there.
Her twitter account is full of her retweeting people supporting her and lambasting Google. I wouldn't be surprised if she gets money out of Google. Worst case the person responsible for firing her is let go.
This is the world we live in. Twitter is a god awful pitchfork paradise. I cannot stand it.
I have to say, this idea that we should assume the best of the large corporate entity to the detriment of the very real human person until "we know more context" strikes me as shockingly and painfully naive.
Is the assumption here that Google is somehow better behaved than other corporations? I don't think there's any factual basis for such thinking. In my experience, corporations are laser focused on money, quarter-by-quarter, with little to no concern for the human cost. Google has been asked for comment and has consistently declined, in my opinion we should give much more weight to the real person here.
We all have much more in common with the AI researcher than the large, faceless corporate entity. If we are at all principled, it's easy to imagine ourselves in a similar situation.
Why the hand-wringing over this email they sent, the suspicion that it was all "hardball negotiating" and "bluffing"? The email to the mailing list pretty clearly outlines their concerns and reaction to Google, it seems a pretty safe assumption that their email to Google management covers much the same information.[0]
This idea that they are bluffing strikes me as insulting: is there any reason to suspect they are bluffing? I don't see that in any of the coverage, by my reading this person felt this back-and-forth over the publication of this paper at the last minute was the last straw, their exasperation and frustration with Google is clear. It seems to me they were entirely willing to resign.
Most people aren't responding because they're defending the actions of a "large corporate entity". They're relating to the people, such as Jeff Dean, that made that decision. A decision that definitely wasn't taken lightly. The people that had to work closely with her, with whom she has publicly ridiculed on Twitter.
She's even tweeted that she suspects there was a team of people behind the decision to accept her resignation. All those people don't put Google first. The hard truth is that she was probably extremely unpleasant to work with it (this is my reading of the above link and from what I've witnessed on Twitter) and they had just reached their limits. Not Google, but the people that had to work with her (I'm clearly not referring to her subordinates who are all apparently "indebted" to her which is a whole other issue onto itself). She comes across as very entitled and quite frankly, although I hate to say it as it's a bit of a cliche at this point, narcissistic.
These are all based on what I've been able to find online. I've never met her in person. I'm trying very hard to not see her negatively but it's also hard finding evidence that's she's not hostile to those that don't 100% accept and agree with her. I'm definitely open to seeing her in a different light if anyone wants to share links. However, just because you research ethics doesn't mean you practice what you preach nor that you're a saint who can do no wrong.
> She's even tweeted that she suspects there was a team of people behind the decision to accept her resignation.
Would that be somehow bad? I'm not even entirely sure of what does that even mean. If I submit a resignation letter, I don't even expect it to be rejected, engineers and researchers are not some kind of ministers. At most some people can consult each others and ponder if its worth to attempt to retain her, but even then the concept is not entirely one of "I/we don't accept your resignation letter"...
I don't think it's bad, I included it to further point out that her resignation wasn't the result of just one person deciding they would prefer she left. But that multiple people agreed that it would be best to accept her resignation. Which in turn, implies many people considered her work conduct inappropriate. If it had just been Jeff Dean, then I may have more sympathy for her case as clearly, that comes across much more personal and unfair, especially if no one else agrees.
The question is whether that Twitter exchange was an isolated outburst (of the kind that reminds us all to stay away from social media, at least using our real name) or an example of typical behavior that led to this firing.
I've never heard of a 300% pay gap between people doing the same job, how does that even happen? Were you paid significantly under par or did that other person somehow hit jackpot? And at what level are we talking about?
I have personally worked somewhere where the person who did the same work as me was paid 3x as me. I was new to the workforce and he was best friends with the CFO for 15 years. Despite my newness and her experience it was the same work.
It's not impossible -- with equity compensation and refreshers, it takes ~ four years to plateau. So if you have someone in the same role for the past four years and compare them to someone just hired, the four year employee may have more than half their income in RSUs while the new hire has maybe 10 percent of income in RSUs. If the stock does particularly well and RSUs grants have impeccable timing, compensation of 3x what you got in year one is a possibility. Especially if a level promotion happens, if you consider that 'the same job.'
Yes, it is a bit absurd. And if you're the first minority hire on a team that hires 1 person a year, you will be paid under the team average for a while, just as if you weren't.
> And at what level are we talking about?
Entry level employees at FAANG mostly. For PhDs managing people, such as the person in the article, I expect there's a much wider band of possibilities.
I'm sorry, but I call bullshit. With salary bands imposing constraints, it's not even possible for you to have made a third of what someone doing "the same job" made.
Ohhh yes you can. Especially when you include equity which is outside the salary band stuff.
You can absolutely get paid that much more. Companies make exceptions all the time and justify two people doing the same job by paying more if someone is more experienced even if the technical skills performed and responsibilities by both people are equal. Instead of being rewarded for working twice the hours to learn something faster due to whatever reason someone can still be paid more for basically accomplishing the same amount of progress in their career but taking more time to do it.
Trust me there are all kinds of reasons you can get paid three times more.
Equity and bonuses. How can you not be familiar with how those work, and just how much they amplify TC and distort bands? Don’t know anyone in finance? You shouldn’t even need to since you’re in tech! Our spreads may not be as bad as theirs, but we can still easily see 3x differences when you take things like sign-on bonuses, target bonuses, initial RSU awards, refreshers, and vesting cliffs into account.
There's no way your coworker doing the exact same job made 3x your salary. I don't care if he was Biden's own son, no one makes that kind of pay disparity doing the exact same work as someone in the same building as them.
Fake news, I'm underrepresented and I never felt like one, I hate playing the minority card and I just focus on being a top performer. Some teams at Google cry all day specially non-eng folks playing the minority card because their work can't be quantified. So they mix up their lack of skills with race BS. I was a SWE and was actually pay very fairly for the work I did, that was my presentation card.
You cannot even say one interpretation is idealism and the other is realism because it is US pop culture bullshit. You notice because your remark has an added indignation.
Have you looked at the self reported salary data? It doesn't really seem to bear that out (although of course any self reported data is going to have issues)
Reading through some of her earlier tweets there seemed to have been conflict around her/ her team pursuing a particular research direction or projects, directions that higher ups were not happy about because of how it might make Google look. This is ultimately the limitation of being a researcher in a company - some topics will be off limits and you’re not allowed to poke around things that will make the company look bad.
By virtue of their research area, her and her team no doubt push against this limit very often. Based on this, I’m not sure it’s possible to really pursue many lines of research inquiry into AI ethics and fairness within a place like Google.
yeah really makes you wonder why these tech companies pretend to be the good guys and hire ethics teams, when they don't intend to follow through when conflicts of interests arise.
Are there any publications or results that show what she accomplished as a research scientist at google? What is considered the leading edge of the field of in AI ethics research? Had she accomplished something of great distinction in this field prior to google? Was her google street view result related to ethics? I would assume that ethics research would involve how an AI can learn to act ethically. Perhaps something about how an AI treats a sentient versus non-sentient impediment, like how it would decide whether to stop for ducklings or just run them over, whether it has the ability to learn a function for long-term social implications of its actions, specifically longer than the training stimulus, which is the remarkable function of the human prefrontal lobe, or only considers short-term task reward.
She had a highly cited paper in 2018 about "Gender Shades", second author. The paper is about dataset bias and the creation of a small size benchmark for bias detection.
Okay; I see. I misunderstood her field of research under the ‘AI Ethics’ umbrella. She’s clearly a leader in the field of algorithmic fairness and social accountability, and this was her work prior to Google, so it seems her criticisms/activism would not have been completely divergent from expectations. It can’t be known without details, but, does it not look like a case of google getting exactly what they hired for?
It's incredibly disappointing to see the first batch of top-level comments here. Timnit is an absolute leader and inspiration in the field, and has been outspoken in dragging the old school of machine-learning-is-money-laundering-for-bias crowd into modernity.
This move makes it clear that Google has no commitment to Ethical AI as a field; they want rhetorical cover for their ongoing bad actions.
According to her, Yann LeCun is a racist just because he said that a specific ML model was biased due to a bias in the data and not because ML learning as a research field was racist.
Lack of insight into structural and systemic bias is indeed the plague of modern ML research, and at this point is so well-understood that carelessness is no longer excusable, and passes into true negligence (or worse).
She didn't call him a racist. She pointed out that there's more to addressing bias than just acknowledging that it's there. If you are going to say "Well, garbage-in, garbage-out!", why do you keep putting the racist garbage in? Why do you, knowing of the problem, keep using biased sources of data, knowing of the potential harms?
That's the deeper question ML research/industry has to grapple with. When these models are deployed increasingly quickly and at scale in ways that can potentially cause massive harm, why is it okay to keep doing the same exact harmful things?
LeCun mentioned that there would have been opposite problem if it were trained on a dataset from Senegal. But why wasn't it trained on a dataset from Senegal? Why do we always see these errors where white-centric datasets produce white-centric results?
It's obvious that while it would be a symmetric situation a vacuum, we do not live in a vacuum. We live in a world with deep sociocultural biases in favor and against various racial groups. And it is unjust to let AI perpetuate and entrench these biases by acting at with this bias at scale.
Acknowledging dataset bias by itself doesn't address the bias meaningfully. In practice, we often treat the bias as an exogenous factor when it is not, moving it outside the scope of our responsibility. But it is very much the product of our work, a reflection of our choices, values, and beliefs about what to prioritize. We can't abdicate our responsibility for it (even if we choose not to prioritize it.)
I think this is a great comment, that would do a lot of people a lot of good to read.
That said, how should a person like Yann LeCun argue his case? Namely that biased models are the result of bad datasets more so than the result of bad algorithms.
I don't think Yann would disagree that biased datasets is a large systemic issue. How should a person like Yann make his point?
It doesn't seem to me that Yann and Timnit disagree all that much. They both agree biased datasets is a problem. They both agree it's a systemic problem. They both agree it does tremendous harm when these biased models are deployed.
I'm very confused as to why there even could be debate between these two people. I cannot find any meaningful difference in their views.
> LeCun mentioned that there would have been opposite problem if it were trained on a dataset from Senegal. But why wasn't it trained on a dataset from Senegal? Why do we always see these errors where white-centric datasets produce white-centric results?
Probably because it was a research AI not a production AI. Having a very diverse dataset at that stage doesn’t help with your research, so it is fine to use whatever is easily available.
At that stage, you are trying to show that your approach can work in some cases. Once you've got that, it is time to expand the research with a wider range of inputs to find out what the limits of your approach are.
For example, if I were trying to make a US English speech to text transcription system, I might start with recordings of assorted NPR programs, because NRP often makes the recording available online along with they transcripts.
That would be great for determining if my basic approach has promise. Once I have determined that, so know that the whole endeavor is not just a waste of time, I could go looking for data that includes speech that has characteristics that would be missing from the NPR data, such as heavy regional accents.
> In practice, we often treat the bias as an exogenous factor when it is not, moving it outside the scope of our responsibility
Thanks the comment and this explication, you clarified the conflict that incident for me.
While I think the sibling comment is correct that LeCun and Gebru would agree on the proximate causes and mitigations of the adverse outcomes of that particular super-resolution model, the issue was (seemingly?) that focusing on the proximate causes can be read as absolving researchers of their responsibility for those outcomes, and avoids discussion of that responsibility in any instance.
Which is a entirely fair criticism of the field as a whole, though it may have been a bit lost in translation to the particular avatars in that instance.
Everyone should read your comment. YLC's initial reply was shockingly reductionist and myopic. How can supposedly smart people, software engineers who are used to thinking in systems and dependencies be so stupid? There is a blindspot in the tech community that is simply terrifying.
“This move makes it clear that Google has no commitment to Ethical AI as a field”
>>
Lol give me a break. You have no idea what was in the email she sent and what she asked for. She gave them an ultimatum. They rejected it. Its at will employment
you live by the sword, you die by the sword. she wants to work for an elite American corporation but thinks she doesn’t have to abide by the rules. well this is not how corporations work and as anyone NOT privileged enough to have only had cushy research jobs can tell you you get fired for insubordination. she can try to re-imagine the system from the outside but who are we kidding? these sanctimonious woke capitalists are no revolutionaries, they just want a piece of the pie without having to incur the concomitant moral costs. ironic that she can expect to marshal support just by ranting on twitter - if that’s not privilege, then what is?
Interesting opinion, since the most recent info to come out of coinbase is that the SJWs were right and upper management are unconscionable assholes. (Oh yeah, let me see now.... they are 'on the spectrum', which is I guess what we white males need to say to excuse our anti-social or racist behavior.)
She is being very selective with the information she is choosing to release. She made demands to Google which they did not accept, but she does not specify what those demands were. She also says that she doesn't have a copy of the post she made to the "Brain women and allies", but she doesn't summarize it either, and given her position, I imagine she could get a copy from someone else in the group if she cared to.
I'd suggest given Timnit is choosing to only tell a very small part of the story here, and Google is obviously not going to publicly comment on an HR matter, especially one this sensitive, that it's best to withhold judgment.
I'm sure she'll make a good asset to any company that wants a totally dysfunctional office culture where everyone is terrified at every moment of doing the slightest thing to upset the wokes.
After reading the mailing list mail in question (pending confirmation that it is authentic) I think Jeff Dean should have been well aware of the situation.
Assuming Dean was aware, this civil discussion should have happened either in the meeting described, or shortly thereafter. So far we only know her story on this meeting, but that story sure makes it sound like there was no discussion to be had.
The comments here are weird right now. They seem overly focused on the spat she and Yann LeCun had a while back when it doesn't seem like this incident is about that?
Hopefully some clarification comes out about what her demands were and the internet drops their pitchfork about something that happened long ago
It's a pattern of behavior. She had a thread just last week calling her employers "a bunch of privileged White men" and complaining that they were disrespecting her by not letting her work on some unspecified research project.
This paints a picture of someone with a history of causing conflict. It points to negative personality traits (narcissism), and these people can be expected to wreck organisations.
Ethical AI is, like many things "ethical" which originated in elite universities, not really all that ethical at all. It's leaders often have a warped ethical framework as well. That's my take on this.
But is this truly the situation where a union makes a difference? She hasn't been fired unfairly for someone who is a senior manager. She has ready employment possibilities and won't experience any hardship.
I'd rather they kept their powder dry to protect employees fired for things like union organisation, and protecting the workforce during large layoffs/reorgs, individuals discriminated against for their race/religion/sexuality, being compelled to take actions that are immoral/illegal etc. None of which seem to apply here.
It could be someone who uses humor as a coping mechanism in stressful situations.
I respond to stressful and/or tragic situations with sardonic/dark/derisive humor. It's a coping mechanism for me. Some people cry. I crack jokes.
Just yesterday I said in a meeting with my boss 'No one wants to touch that project, because the customer gives out negative feedback like cheap halloween candy.'
If I were fired from my job in a way I felt was unfair, I'd probably do something similar.
> Apparently my manager’s manager sent an email my direct reports saying she accepted my resignation. I hadn’t resigned—I had asked for simple conditions first and said I would respond when I’m back from vacation. But I guess she decided for me :) that’s the lawyer speak.
> 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. Then she sent an email to my direct reports saying she has accepted my resignation. So that is google for you folks. You saw it happen right here.
In the email she posted, it seems that they "accelerated the timeline" for her termination because "certain aspects of the email you sent last night to non-management employees in the brain group reflect behavior that is inconsistent with the expectations of a Google manager", i.e. a separate email from her conditions email.
It seems like she was ready to leave if they didn't meet her conditions, and that this would have been a reasonable outcome for her. However, it bothers me that they wouldn't even come to the table about the non-condition email.
Like OK, "inconsistent with the expectations of a Google manager". That's some weak shit. It had better be something pretty fucking severe for them to fire her immediately and not, I don't know, do some planning around transitioning her projects? Work out what to do with the team she herself managed? Be given the grace of leaving on a good note that an engineer of her caliber deserves?
She's a leader of their ethical AI group. What happens to that group and all of their research and direction now? Who do they hire to replace her? This isn't the same level of impact as firing a cog in the wheel IC who gets too uppity with organizing.
This stinks like a convenient excuse to shoo an AI ethics researcher asking some inconvenient questions out the door.
To me, it stinks like a convenient excuse to remove someone who was difficult to work with. Only last week she slandered her bosses on Twitter, which is a public forum.
Matthew Garrett, a senior security researcher at Google, is very critical about Google on his Twitter all the time. As he has pointed out, he's still very employed.
Neither of us are on her team at Google. (I'm certainly not and I doubt you are.) We have no idea how easy it is to work with her.
Certainly if I were working with an AI ethics researcher I would expect them to ask difficult and uncomfortable questions. I think that's part of the job description.
Because he was provinding real value. Her job was mostly a PR play by google, and she went contrary to that by actually slandering google all the time and stiring tensions...
Google historically tolerates generic criticism of the firm or its actions. Gebru routinely makes slanderous and offensive allegations against specific individuals, not Google as a whole.
That said, Garrett makes clear his ideological sympathies on Twitter. Undoubtably comments like this one don't endear him to his management chain:
When @computerfemme was fired, the VP of security at Google Cloud sent out an email to the security organisation containing one absolutely false statement and one misleading statement, both of which portrayed her acts in a more negative light.
Pretty serious stuff. He doesn't elaborate on what those claims were. The woman in question was fired from the security team for violating IT security policies: in her words, "All I did was make a popup to share the labor notice Google has to share with its workers", something that Google seems to have recurrent problems with staff doing. In other words she was rewriting the contents of web pages to meet her own political objectives.
If I were the VP of Security at Google I would have serious concerns at that point about insider threats from sympathisers on the security team. You can't be publicly slamming your own VP as dishonest for firing someone who subverted IT policies if you actually work in security.
Let us pretend together that you are the VP of security at Google. I’d suggest you take a stroll through Matthew’s twitter page for the past year or so and mark down everything you find seriously concerning and in your view making him incapable of performing his job. You might find quite a bit, and you might think “I can’t believe the real VP of security at Google hasn’t fired him yet!”
And yet, somehow, he’s not enough of an “insider threat” to be dismissed.
Well did he email his management with a list of demands they weren't willing to meet, saying that he'd resign if the demands weren't met? Because that appears to be what Gebru has done.
Also, did he publicly attack his colleagues on Twitter for their race? “Bunch of privileged white guys” is not something you want to read from someone who you have to work with every day, she should have been fired immediately for that alone.
> The woman in question was fired from the security team for violating IT security policies: in her words, "All I did was make a popup to share the labor notice Google has to share with its workers", something that Google seems to have recurrent problems with staff doing. In other words she was rewriting the contents of web pages to meet her own political objectives.
This paragraph that you just wrote is one of those claims that are wrong and/or misleading
How? That quote comes from the person who was fired. She explained what she did because she's proud of it: she injected some code into a web page about unions and organising telling Googlers they had a right to organise.
The Google security team is frankly deeply worrying at this point, and I say that as someone who used to work there on security related topics (but consumer account security, not internal). A small number of them have privileged access to mandatory Chrome extensions that are used for various security objectives, yet they seem to have no qualms about abusing them to advance their own political agendas. Moreover they don't recognise that this is a problem.
Controlling Chrome extensions that can inject JS into arbitrary websites equals root at google.com because everything they do is admin'd via web UIs.
What if you're a law firm who advises companies faced with union action? Is it safe for you to use GSuite or whatever it's called today? My guess is no, because the people who ultimately control Google's IT infrastructure are (a) willing to break the most basic of IT security rules and then boast about it publicly, and (b) many of them are politically extreme by the standards of most of the world.
Most astoundingly of all, I don't believe this is the first time Google has fired people for injecting JS into websites using internal security mechanisms to advance political agendas. In any normal company there'd be a mile of controls, processes and guard dogs surrounding any mechanism that could do this by now as it's been abused before. Apparently Google Security do not have their act together on this.
> "All I did was make a popup to share the labor notice Google has to share with its workers"
Does not mean rewriting the content of webpages, or having "root at google.com". It's better if you'd stop such speculation.
The reality is much more pedestrian than what you describe: the person that was retaliated against only wrote simple, inert, strings of plain text.
That text was not injected anywhere, and there was no arbitrary code involved.
In fact, everything was safe, and went through the usual security controls.
The only "danger" in having that, is that people can read it.
Ultimately, it's laughable how you think that informing other coworkers about the NLRB is some kind of politically extreme agenda. It's clear that you have an axe to grind
It went through the usual security controls yet she was immediately fired for doing it? Really? Do you understand how that sounds?
Please do explain how someone makes a message appear in your browser when you visit a particular website, without having access to a highly privileged browser extension? That can at minimum monitor the URLs you're visiting?
I honestly really hope you aren't on the Google security team, because you seem to be missing the issue here. Someone who had a high level of access was fired for violating basic trust, apparently with the collaboration of her coworkers. What other things could this person have done to advance her political goals by subverting her access for things it wasn't meant for? Outsiders can't know. All they can see is that the company doesn't have a grip on its own workforce. That is not confidence inspiring.
> it went through the usual security controls yet she was immediately fired for doing it? Really? Do you understand how that sounds?
Yes, that's crazy... And that's exactly why people have been angry at the retaliation
> Please do explain...
To get more details, the appropriate thing to do is to draft a post that explains how these things are accomplished at Google and get it approved for publication by the legal review team. I've never done it and, while I don't exclude that I might do it, it probably won't happen soon.
Argh, being fired for this isn't crazy. What kind of "security control" allows changes to security systems that aren't for the purposes of improving security? Whatever controls you're talking about are clearly not controls of the sort a normal auditor would recognise.
The white men, where she used both their skin colour and gender in a derogatory manner, were accused of trying to shutdown minorities. I would be very surprised if those men did anything with an intent to squash minorities. They were most likely making business decisions to further Google's intent.
Email claims she was told to stop publishing a paper with basically zero reason/conversation matching the tweet.
Hiring has apparently failed to meet goals with zero consequences suggesting a lack of concern with minorities.
I don't think someone misinterpreting disregard for ethical ai research as disregard for minorities is slanderous, heck it might be charitable as the later is known.
"Who do they hire to replace her?". That is an excellent question, and a good time to remember that Identity Politics != Ethics. Most big ethical questions in AI have nothing to do with race or gender, instead affect us all. Some say AI is an existential risk for Humanity at large.
The casualness to make this now about race/gender as well when that wasn't even her inital argument.
I can't imagine working in this kind of environment. It must be exhausting, always having to tread on eggshells or end up on a twitter firing squad. Not being able to argue points, without it ending up being about race/gender.
Do we really have to pretend that someone blessed to be working on AI at google with the greatest minds on something so exciting is disadvantaged compared to the rest of us?
I remember reading on her twitter around the Yann LeCun scandal time, she acknowledged her privilege while calling Yann on his. I didn't like how a man's output was seen as the result of privilege instead of hard work.
[Redacted, since I don't want to fuel speculation]
It always was (also) about race/gender!
That said, I don't know Timnit, and I don't know her specific circumstances... I believe that Google is full of people who wants to solve these problems, but somehow leadership fails to capitalize on that. It might only be a few bad apples in the leadership, but each one wields a huge amount of power (hiring and firing people)...
I mean, I don't know what "dared" has to do with it. It's unfortunate that we're working off incomplete information here, but that's not our fault - unless there's a recent leak I haven't seen, nobody's made the email available to read.
Typically they're numeric diversity hiring goals, something like "40% of our new hires should be women" or "10% of our new hires should be black". You sometimes see more recruiting-targeted ones (like Gitlab's pledge that 95% of its reachouts would be to underrepresented groups), or less numeric things like efforts to build employee affinity groups.
Diversity requirements are a common practice for large law firms.
Many Amlaw 500 firms have agreed to the Mansfield rule[0].
> Now in its third iteration, the Mansfield Rule Certification measures whether law firms have affirmatively considered at least 30 percent women, attorneys of color, LGBTQ+ and lawyers with disabilities for leadership and governance roles, equity partner promotions, formal client pitch opportunities, and senior lateral positions.
Mansfield rule certification is independently audited on a biannual basis.
It's also common for clients to have their own diversity requirements[1].
I actually find it quite unethical and potentially rather manipulative to share with your future employees that you fought tooth and nail for them. Essentially implying that they wouldn't be there if it weren't for you and immediately setting it up so that they feel indebted to you... it's chilling to see that tweet tbh
I was curious so I looked it up. According to Timnit Gebru's Wikipedia page she was born in Ethiopia but her parents are from Eritrea. I can't find it again but I saw something somewhere about "Timnit" meaning "wish" in Tigrinya, a language spoken in Eritrea and northern Ethiopia.
Unfamiliar with a culture, it is easy to make these mistakes. I thought Taylor, Blake, Charlie, and Hayden were male names. Apparently they're practically unisex in their use.
There's so much context missing here. What did she post on Google Brain? What were the demands she made to Google? Are these in the public domain?
(If they're not public then I imagine they'll be leaked eventually if this gets enough media attention. Are we witnessing the opening act of Google's next enormous Damore-style PR clusterfuck? Grab your popcorn.)
I personally have ever seen the 'do x or I quit' move work exactly once. The second time they tried it they took them up on the offer. They were even very nice to work with and were part of the 'do not fire this person or it is chaos' team. When he did it the first time he said 'I can only do this once if I try again they will fire me on the spot'. He did it a second time just to see if they would match his external offer (they let him ride the 2 weeks at least). You usually only get one shot at that sort of move. If you do it all the time, managers start looking for ways to offload you. You will be seen as a threat to org stability as you come off as unreliable. No matter how key you are.
The comments here are disheartening, essentially a combination of “It was Google’s right to fire her” and “She seemed ‘difficult.’” For an industry-operated “Ethics in AI” group to have any teeth, it can’t fire people when they do research that makes them look bad, or when members of the group are ‘difficult.’ (Multiple members of the team she managed have said that she was a great manager and this came as a total surprise [1, 2, 3, 4, 5]; so it’s also worth asking who she was causing difficulties for... and whether it was her, or just her research.) Google can’t have it both ways: it can’t credibly claim to be home to meaningful research about AI ethics, but not provide the tenure-like academic freedom protections that enable researchers in universities to discover and publicize results without regard for whether they paint a company’s products or direction in a bad light.
people here have been passing premature judgement, as have you. You are projecting criticism of the commenters onto google; it hasn't been stablished that google fired her because she was difficult or made them look bad. There is indeed a scenario where that is the case, but there are others where Google is right to fire her without undermining their ethical position; to say otherwise is to say that she is infallible, an attribute usually reserved to religious groups. If she wants some non faith based support, she should be transparent about the terms of the email where she put forward conditions to the company.
Seems to be a large number of people defending Google on here when every other other day, Google is the target (and rightly so) for criticism on HN. Would this be a bigger issue with the HN crowd had it not been a black woman?
Internal ethics commissions aren't independent and cannot really work in opposition, so the effectiveness is always questionable and it doesn't seem that ethical behavior of Google products was the issue in this conflict.
AIs were accused of being biased so Google may have hired her as a defense mechanism. Her job was to justify Googles actions, basically a PR job and not a control organ.
Dunno where you've worked but people who are disruptive to team functioning can (and should!) be terminated. Nobody should have to put up with a hostile work environment.
Lots of people have jobs where they should be disruptive to other teams. Financial audits are disruptive to accounting. Legal can disrupt just about any team. HR can disrupt hiring decisions. Security disrupts my desire to move fast and break things.
That does not create a hostile work environment. It's healthy and normal. I would expect an ethicist to fall under a similar category.
In my opinion, she is being very unfair to a lot of her colleagues by not telling the whole truth and at the same time making very serious accusations. She sent email to large group of employees which violated some standards expected from a manager by Google. She is giving all the information except for what exactly she did. It's ironic that she is in "ethics business" but at the same time not realizing that it's unethical to withhold what she did while launching attacks on everyone.
> but at the same time not realizing that it's unethical to withhold what she did
Have you considered that she has to remain vague due to the potential litigation coming down the road? In fact didn't she leave some tweets hinting at that ? ("Everything I say will be used against me ..." ?)
I can't imagine a reason that she'd be free to say Google fired her because of her demands, free to post the termination email, but not free to say what her demands were.
From what I understand, her access to said information (I assume you mean the content of the email she sent to the group) was cut off, so she wasn't actually in a position to (legally) share it any longer.
This seems to be it though, and I quoted some (I think) relevant parts of what she's criticizing:
(1) If she was difficult with/for everyone then she would have been fired earlier. It makes sense for the people who liked her management style to be supportive and vocal about her, and those that were not happy to not say a thing. Especially in the cancel-culture that we now live in and the economic-uncertainty.
(2) I don't understand how someone can make public accusations without providing the full picture. Maybe she is legally constrained, but then she shouldn't have said anything and handle the whole situation legally first and then write what she wants to write about it. Google might be wrong here, but we definitely cannot see that.
P.S. It seems to me that the only person who handled "firing" professionally was the most ridiculously dressed person on the planet who goes by the pseudonym Dr. Disrespect. So much drama going on around nowadays.
As somebody who has been involved in various diversity efforts for years, I'm honestly not that upset about all of this. There are real problems associated with discrimination in the US, for example in the criminal justice system. Likewise, there is discrimination in less high-stakes areas like tech.
However, the main problem with Timnit and her colleagues is that they come into every situation assuming racism even when the situation is ambiguous. Likewise, if you disagree with her or want to have a nuanced discussion, she'll accuse you of imposing intellectual labor on her and in some sense you are racist for forcing her to engage in discussion.
Again, so while I appreciate how she is purportedly working to make my life better in tech, there needs to be a real awakening as far as giving people the benefit of the doubt and allowing for discussion without shaming. Google has accumulated many people with this type of attitude who have created a culture of intimidation. You can see all of this happening in real time as her colleagues come to her defense knowing little about the situation and assuming that she was fired because she was black.
I think this mindset is especially toxic when people have access to data accumulated by Google, nor would I trust them to handle user data in confidence. This is not a small problem.
308 comments
[ 1.7 ms ] story [ 331 ms ] threadTrying to get context, I see big names professors from Stanford, Caltech, Berkeley, Cornell all commenting/retweeting on how bad this is, so it's a big deal for sure
> I was fired by @JeffDean for my email to Brain women and Allies.
In a later tweet, says she would repost the email publicly but she no longer has access to it. I'm guessing it will leak soon.
Google Brain Women and Allies is a mailing list within the team.
I don't work at Google; I work at Amazon. I can only speak about my experience, but it's likely the same at Google.
You don't have to be a part of an affinity group to be a member of the mailing list (that's where the ..and Allies portion comes into play).
The 'Google Brain Women and Allies' mailing list is probably full of women from the Google Brain team, non-binary and men from in from the Google Brain team, and others who aren't on the Google Brain team but are interested in its work.
Again, I don't know. But this is educated conjecture based on my experience at another FAANG.
At the hazard of becoming 'political' and triggering people, affinity groups are typically created for groups that have traditionally been underserved, marginalized, or have special considerations.
As a cis male, I don't really see the need for a 'Google Brain Men and Allies' distro list because the issues that men face in the workplace are generally the same issues everyone faces. Women (and nonbinary people) can be confronted with issues that men simply aren't.
---
At Amazon, there are groups for Black employees, Latinx employees, LGBTQA+ employees, transgender employees, women, disabled employees, etc.
Sometimes these groups talk about work; but sometimes they talk about just life things. For example, often cisgendered employees will have children that come out as trans. So they'll reach out to the transgender community for help and support as to how to be the best advocate for their child.
Or, as a personal example, I moved to a new city to take the job at Amazon. I was looking for a therapist, and it was important that the therapist be supportive of LGBTQA+ issues and ethical non-monogamy. So I reached out to the LGBTQA+ group for therapist recommendations.
But when I hear about a 'Google Brain Women and Allies' distro, I'm more thinking about peers which discuss things which impact them. I wouldn't get too hung up about the 'Google Brain' portion of the distro; it's likely just a way to self-select a group of peers which share a common interest.
Gender equality activism has been part of US politics since at least 1776: http://www.thelizlibrary.org/suffrage/abigail.htm (Abigail Adams advocating to the constitutional convention)
Is there an affinity group for the overwhelming majority of "Latinx" people who don't use or like this term?
In my opinion, this leads to division and contempt/hatred for the other group. I live in a super progressive US state, and the university near me got a lot of backlash for hosting White affinity groups.
Does anyone else remember when she picked a fight with Yann LeCunn by deliberately misinterpreting his remarks to cast him as a racist?
Seems reasonable to assume she picked another fight. Reading the Twitter threat, it reads like she made demands in a way visible to a large number of employees (something about authorship on a paper, to skim the Twitter thread) and perhaps attacked Google while being a manager there (the emphasis on talking to non-management employees sounds like this). I assume the email will leak, this being Google we are talking about.
LeCunn made repeated attempts to explain his position, only to be virtually shouted over about he wasn’t listening to people of color because he pointed out that a model trained on White people learned to produce images of White people, and that if it had been trained on Black people it would produce images of Black people. It was a research paper, not a production system, and Gebru and her acolytes were much more interested in scoring cheap points than in having a serious conversation about either ML fairness or the merits of this particular paper.
Seriously, the primary sources can be read by anyone.
But this claim is certainly not necessarily correct, and he shouldn't have been so confident about it. Any part of a system can contribute to bias and that includes the model design, not just the data. If they want the system to work then they need to actually demonstrate it works, not just say it could with X change without testing that.
Though I remember people arguing it didn't work properly at the same time they were saying it was evil (because it contributed to surveillance). Which is odd because if you don't want it to exist, you shouldn't want it to work either.
The most obvious and easiest way to clear this up would be for the original researches to train the model on a "Senegal" dataset as mentioned by LeCunn and see what the results are.
While it would be a symmetric situation a vacuum, we do not live in a vacuum. And acknowledging dataset bias by itself doesn't address the bias meaningfully. In practice, we often treat the bias as an exogenous factor when it is not, moving it outside the scope of our responsibility. But it is very much the product of our work, a reflection of our choices, values, and beliefs about what to prioritize. We can't abdicate our responsibility for it; the stakes are too high.
I am not sure why the wording around this has to be so abstract.
This would be called “research”. That is what she was hired to do at Google. For everybody of every race ethnicity orientation able status and background that would be honored to give their best effort at the opportunity to do research at google, acceptance of resignation was the right thing.
My only point around her feud with LeCunn is that she threw a fit because people including LeCunn pointed out that it is not necessarily a race thing, and she wanted to give it a race spin without confirming that it was actually the case.
And I am not at all commenting on all this drama around her resignation/being fired without getting a look at the contents of the two emails: the one she sent to Brain Women and the one with her demands. But I know this: if I were to issue an ultimatum to my employer, I am treading dangerous waters and should be able to digest the outcome of that, including being fired.
This is not surprising. Black faces and White faces are not the same data manifold. This is like training a network to upsample oranges, then running it on an apple and being surprised when the result is an unusual color and texture for an apple.
I’m not sure what else there is possibly wrong here. Is the width of their convolutions racist? Their choice to work on super resolution? The fact that they released their work for reproducibility?
Not that parameter specifically, but if you say the data is wrong then that means any other parameter might be wrong too. More data might mean the model is too small to fit it, or the hyperparameters might be wrong to train it, or you now have the wrong ratio of other phenotypes (let's say that instead of races…) in the training set and their results regress.
Also, if you're adding people with darker skin, that of course means the pixel values are lower. That matters for image processing code, things like SAD thresholds or noise reduction will work differently.
> Their choice to work on super resolution?
Superresolution is only useful as a toy and they should have mentioned that when they put up a live colab, yes. They added a disclaimer later on and it was good - there was a big issue where people were convinced this was somehow a surveillance technology because they watched too many TV shows, even though it literally can't work that way!
in particular because features contrast (dynamic range) is lower for darker faces.
>Is the width of their convolutions racist?
Not width. As a result of the above mentioned lower contrast, the racist here is the sensitivity of the resulting Gabor filters produced by the training in the first convolution layers and the Gauss filters in the next layer's. I suspect to deal with that problem one would have to normalize dynamic range of the faces, i.e. it would look something like kind of lightening of the dark faces and/or darkening of the white ones.
Since everyone else is just saying "you don't get it" without explaining what "it" is, I will provide some brief avenues of exploration. Because they are brief, they will be coarse and imperfect, and aiming to give you directional assistance on the topic.
So with that massive preamble because I'm not seeking to argue, here you go:
* Pre-trained models already encode much of this bias and if you don't use them you won't get very far very fast
* Large available datasets also reflect these biases
* AI model performance is generally measured against this dataset, further entrenching the bias. If you're better on Indian face generation it will give you barely any benefit on most scoring methods
* Saying "the outcome is only biased because the data is biased" misses the fact that the performance is on the biased data
There's a little more to it but I believe that's the meat of it. Anyway, not too keen on arguing this. Just sharing because it took me some work to figure out what they were talking about and I wish someone had explained it to me so doing so here.
For instance, if I were to champion metric A which purports to measure performance on human faces but it really only rewards performance on Senegalese then models that do better in general may not be recognized for being better.
In an isolated sense this is not a problem. However if the mainstream is that all metrics that are taken seriously are dataset-biased then we'll have an environment where the models will be trained on biased datasets in order to be successful.
For instance if everyone uses LFW to determine how good facial recognition is, then Senegalese fine-tuned facial recognition tech will not be recognized as being good at facial recognition.
So the argument is that dataset bias is built-in to our approach to the problem. I, personally, think that this isn't malice. We need benchmarks to judge approaches against each other. Benchmarks always have a first mover advantage and a massive path dependence issue. The first benchmarks aiming for things on humans do not reflect humanity accurately. These benchmarks became standard among the community. To be taken seriously you have to do well on benchmarks that are standard in the community. Dataset bias is then natural in newer approaches because the approaches are judged against how good they are against the inaccurate (if you will) benchmarks.
So no one need be racist or anything for the end result of the field to end up being discriminatory.
I don't think a successful approach is to call people racist over this. After all, it isn't malice that guides them. The discrimination comes from the sort of historical accident that has North facing up on a map. And no individual is really racist. It's sort of like the Bechdel test: no movie is crappy simply because it doesn't have two girls talking to each other, but if very few movies have two girls talking to each other about something other than boys, then it makes you think "hmmm, why's that the case".
That's, like, not even culture war, it's just basic correctness of the reference datasets. If a fruit classifier was missing oranges, we'd just fix it and move on.
Like, for instance, HN has people who will bring up privacy violations of big tech constantly. They see their role as making sure the conversation is happening. Not justifying this. Just aiming to understand it.
For my part, I prefer to take the approach you're talking about because I, too, think that the fastest path to this is getting the photos, labeling the photos, and then lobbying for inclusion. Ultimately, I think it's okay if things optimize fast for growth and then we fix up issues afterwards. So the people building the benchmark sets weren't able to get a set that's representative of humanity. Should they have waited till they could have done that? IMHO, no. Rapid release moves the state of the art forward and then we can put in all of these corrections as we move.
Then there's the question of whether all-humans dataset is a good thing or if instead a thing that is white-humans and another that is black-humans is better. Anyway, all said, my personal approach to this problem (if I cared about it a lot, which I don't) would be to say "Current benchmarks and training data available bias towards certain races. I'd like to build X/Y to solve that. Here's what I have so far" etc. etc. I think positive engagement like that yields better results because the vast majority of scientists actually aren't weird race supremacists and the vast majority of AI researchers will gobble up any more data you give them which is segmented and labeled differently, the greedy bastards :D
Some challenges that I can still see:
* Getting the data. Might not actually exist.
* Labeling the data. Probably needs some work.
* Getting it into the benchmarks. It'll invalidate old scores, so there's just a product adoption problem here. I don't know how the community handles newer releases.
As a last aside, I suspect that this conversation ended up the way it did because:
a. It's charged. It's race-based differing outcomes. That's a sensitive subject.
b. People feel unheard. This is natural. Like, this is not an 'interesting' problem. It's literally just a data error so the luminaries in the techniques part of the field aren't really that interested in it. And the techniques part is where the sexy is.
c. This sort of thing has a tendency to escalate. One side says "You're not listening to what I say" and the other side says "I'm not racist. I don't get why you're calling me that" and before you know it it becomes "You have to be racist to be ignoring me" and whatnot and de-escalation becomes impossible. Especially because everyone rewards the loudest on each side.
Honestly, I think it's quite interesting to observe and to understand as just a view into the human condition but we use AI models professionally in the GIS space and professionally we just don't go near this at all. No part of me finds it interesting to solve or to interact with the discussion in any way. I only sort of participated in this here because I think I managed some insight into what it is and I wanted to write that down because I wish someone else could have accelerated me into it.
Anyway, I think that's all the insight I have on ...
Do something productive and uncontroversial, or something unproductive and very controversial?
Am I missing anything?
I think it’s clear Gebru was acting in bad faith and LeCunn was baffled and trying to deescalate. Gebru responds to each attempt to deescalate by a reply designed to further rile up Twitter, without actually talking to any of the points LeCunn makes. Virtually all of her replies are some variant of “you are wrong, but I won’t say why” and “listen to me because I am Black”.
Edit: incorrectly claimed the spat was before 2020, it wasn't. 2020 just feels like an incredibly long year.
Huge, HUGE [citation needed].
The racism here is in the behaviour of treating some people specially because they are of "wanted" races or genders. I'm criticising racism, not engaging in it. But clearly a lot of people especially in the US are so far gone they'll never recognise it.
The same is true of AI research Twitter.
I'd say she deserved to be fired for she is a racist who's damning to a normal society.
Whom did Gebru discriminate against, based on their race? No one.
That episode is unfortunate imho, but she only suspected racist rationales... Calling out racism does not make you a racist, just like calling something "fishy" does not make you grow fins and gills
Which she doesn't.
Most notably, people here have been raising concern about the discussion with Yann Lecun, but that was also just a civil, if tense, discussion.
You're better off not twisting the meaning of words to favor your argument, though. It'll just create more problems down the line
And her paper was rejected and her immediate reaction is to demand the company to reveal the identity of every reviewer? Yeah, right.
> Virtually all of her replies are some variant of “you are wrong, but I won’t say why” and “listen to me because I am Black”.
I don't see that in the summary you posted.
Please don't create accounts to break HN's guidelines with!
https://news.ycombinator.com/newsguidelines.html
(BTW This is actually the fun part of the Damore story: the guy think he is smart by explaining how "evolution" makes gender different but misses the ABSOLUTE main point of evolution which is that "to survive to an environement you must fit in it". He didn't fit in the google environement, he got fired.)
Be weary of "rockstar" scientists. Science is for the most part about incremental insights, and shouldn't make you famous.
Without knowing more context, it sounds like she tried to do some hardball negotiating by threatening to leave, and got her bluff called.
Quoting the tweets for posterity, link at the bottom:
> I need to be very careful what I say so let me be clear. They can come after me. No one told me that I was fired. You know legal speak, given that we're seeing who we're dealing with. This is the exact email I received from Megan who reports to Jeff
> Who I can't imagine would do this without consulting and clearing with him of course. So this is what is written in the email:
> Thanks for making your conditions clear. We cannot agree to #1 and #2 as you are requesting. We respect your decision to leave Google as a result, and we are accepting your resignation.
> However, we believe the end of your employment should happen faster than your email reflects because certain aspects of the email you sent last night to non-management employees in the brain group reflect behavior that is inconsistent with the expectations of a Google manager.
> As a result, we are accepting your resignation immediately, effective today. We will send your final paycheck to your address in Workday. When you return from your vacation, PeopleOps will reach out to you to coordinate the return of Google devices and assets.
[1]: https://twitter.com/timnitGebru/status/1334364732480958467
You cannot fire someone for membership in a protected class (race, religion, sexual orientation, etc) or participation in protected activities like reporting an employer’s violations of the law, but it’s difficult to prove intent in such cases.
There are exceptions for unionized workplaces with collective bargaining agreements. Employers in those situations cannot fire employees without “just cause.” Google R&D is not unionized, however.
https://en.wikipedia.org/wiki/At-will_employment
I assume they wanted her out and she accidentally gave them immunity, so they jumped on it.
She made some pretty serious demands, not having been met, with such politicization there's no reasonable way to accommodate a longer handoff, and she knew that. Waaay too much risk there.
She set her own terms, they accepted, that's it.
She should have been more transparent in this, because in her Tweets she wasn't forthcoming that it was her terms.
There are a ton of people who are sympathetic to those who are oppressed, but misrepresentations, cancelling etc. on the wrong terms are going to lose general sympathy very quickly.
None of this should be construed as a measure of whether or not Google is 'Good or Bad' or whatever, all these things are individual situations, with their own unique circumstances.
All we know is that she gave a hardball ultimatum, and Google accepted option 2. I don’t know enough about what’s going on, but if you say “Do this, or I quit,” you have to be prepared for them to show you the door immediately. That’s how at-will employment works.
"“Do this, or I quit,” you have to be prepared for them to show you the door immediately. That’s how at-will employment works."
That's pretty much it.
She must be smart enough to know better and possibly she was looking for an exit of some kind.
> In the letter she criticised the use of pre trained language models in Google's products (e.g., BERT is now used for most searches, machine translation, etc.). Apparently, the two conditions she mentioned concern how Google goes forward in deploying these models despite the warning of (their own) AI ethics researchers about biases that are manifested in these models.
Is it though? There is bias in everything.
There is bias in Googles original search - it crawled content. That content will be biased.
Google search has been biased since the start. It has nothing to do with AI.
Biased is also contextual: what does it mean for all of us non-Americans to see tons of American content in everything - the 'American bias' is overwhelmingly the strongest bias, where are the concerns about that?
And how does bias imply a 'lack of ethics'?
The entire consideration is ridiculous:
1) AI is not special and odes not deserve it's own ethical oversight. Every social tech has issues and it all needs to be thought about.
2) Someone with no real training in the issue may very well merely be injecting their own politics into the situation, and possibly overstating issuses.
3) There's nothing objective about morality or ethnics, so it's really hard to even find such at thing as 'objective'.
What the company needs is a clean, comprehensive framework for the issue, and probably some independent oversight that can given them private assessments of where there are red flags.
Not individuals who want to make a name for themselves on the issue publicly, and who might be part of some kind of ideological movement.
Your disagreement seems less to do with that and more to do that you disagree with where her ethics rest. I think the reason why she pushes for marginalized groups is that those are the groups that are less likely to be able to stand for themselves and thus become the victims of the bias. A bias against corporate or powerful interests will likely be rectified relatively quickly.
But hiring activists, and particularly those with a focus on a very specific technology ... I think is a double-wrong.
'Ethics' is all encompassing, every company has to deal with it ... it needs probably specialization and some cold calculating.
If there were some gigantic loophole whereupon marginalized people were truly thrown under the bus here, then maybe there's a case, but I seriously doubt that.
It's also pretty common in roles like sales and PR to be shown the door the day you give notice, especially if you're going to a competitor.
If you make an ultimatum to your employer, and they don't want to accept, assume that it's over, right then and there.
Unless you're Michael Jordan, or have some kind of really existential terms with the company.
But at the very least, I'd guess that this will make any potential future employer think twice if they really want to be her manager.
And lastly, I remember the name Jeff Dean from reading about advanced AI research, so if someone publicly says "Jeff Bean says XY is unbearable", I'd be inclined to just believe it.
All in all, this looks to me like Timnit Gebru could improve her career chances by being a bit more diplomatic with her public announcements.
He's also a rather mild mannered and very nice man. When I worked there I never heard anyone say anything bad about Dean. He rarely expressed strong opinions even on engineering topics; the sort of guy who prefers to win debates by simply coding up the best solution.
The name Timnit Gebru also rings bells for me. Last time this name came up she was busy chasing Yann LeCun off Twitter with pitchforks. LeCun! One of the other major luminaries of AI research. In fact she stirred up so much trouble and made so many outrageous accusations he quit Twitter altogether. She appears to be a serious troublemaker, far more interested in identity politics than AI. She was an embarrassment to Google before and it's good for them they finally canned her, but seriously, why did it take this long? And who will hire her next? I'm sure someone will be dumb enough to do that but hardly any companies have openings for "ethical AI researchers", and she already burned her bridges at Facebook.
Because public division creates sides, and taking one of those sides gives you favour with one group.
A pragmatic, mild mannered person who works on such issues behind the scenes is not going to be popular.
But division makes enemies in one camp, allies in the other.
Now we all know someone's name whereas we wouldn't have before.
So one can make a career out championing said allies.
I think a lot of people have this instinct, just expressed in different ways.
that just reminded how Marisa resigned - if i remember correctly she just called on Monday morning to notify that she is already at Yahoo. So the "at-will" works both way.
>> certain aspects of the email you sent last night to non-management employees in the brain group reflect behavior that is inconsistent with the expectations of a Google manager.
as my friends managers explained to me - a manager represents the company to employees, i.e. when a manager speaks it is basically the company speaks and thus [until the company reacts quickly by disowning the manager's words] the company bears the burdens and responsibilities that the manager made promises/representations about - similar to that relationship between Pope and God pictured in the Dogma - so managers are trained to be very accurate with their words, and understandably a manager going off-rails is like a fire emergency to be dealt with immediately.
So looks like Timnit sent an e-mail to the group which wasn't OK in the eyes of the higher management, then sent a strong mail about her demands to the same management and used "I'll resign otherwise" card to force her way.
The management, agitated from the first e-mail, didn't buy it and sent her home. The way they did is neither elegant nor polite but, she pushed a lot to get this response it seems.
This is my understanding of the issue and I may be completely wrong but, this is what I was able to decipher.
Or she was on her way out anyway (or at least ambivalent about staying, it sounds like) and wanted to provoke Google to fire her. These days getting fired as a female minority is a good way to elevate your reputation among a certain crowd, maybe get a more lucrative offer elsewhere, and possibly get a payday if your employer didn’t have good lawyers.
Just look at her Twitter mentions right now, she’s getting an enormous amount of attention, including job offers and lawyer recommendations.
This is the olympics of mental gymnastics.
https://www.platformer.news/p/the-withering-email-that-got-a...
It also sounds like she did lawyer-rammed the Google a year before, so they were tip-topping around her strictly by the book, and with such clear an offense as that email she just threw the HR a freebie, as another commenter put it, so they could now get rid of her strictly by the book.
> Have you ever heard of someone getting “feedback” on a paper through a privileged and confidential document to HR? Does that sound like a standard procedure to you or does it just happen to people like me who are constantly dehumanized?
Taking a dispute with your management chain re: feedback that said management chain felt they had to take actions like a private, hr-involved feedback session, and broadcasting that dispute to a mailing list of peers, downlevels, etc.
> But now there’s an additional layer saying any privileged person can decide that they don’t want your paper out with zero conversation.
This was, from the perspective of google management, probably very much not a reasonable characterization of the aforementioned circumstances. Managers represent the company, including publicly. Timnit was obviously in an awkward position where she seems to have been a player coach, ie a manager also doing IC level work (in this case, research). It feels like she expected her actions as an IC not to be viewed as actions that also came from a manager. That's a very hard dual role to have.
Third, maybe shouldn't get you fired in a perfect world, but prolly will in ours:
Public internal criticism of DEI efforts on the part of your management chain. Imagine how that reads to a line-level employee at Google. Particularly an URM.
Fourth, same vein -- auguring on an internal mailing list to get the CBC to criticize your employer
> I believe that the Congressional Black Caucus is the entity that started forcing tech companies to report their diversity numbers
I'm only paying attention because I'm curious if Timnit was hired to -- if you will excuse the phrase -- whitewash google's policies, and she didn't understand what Google was buying; or if there genuinely was some conflict here. I suspect both from different players in the executive ranks.
This is the world we live in. Twitter is a god awful pitchfork paradise. I cannot stand it.
Is the assumption here that Google is somehow better behaved than other corporations? I don't think there's any factual basis for such thinking. In my experience, corporations are laser focused on money, quarter-by-quarter, with little to no concern for the human cost. Google has been asked for comment and has consistently declined, in my opinion we should give much more weight to the real person here.
We all have much more in common with the AI researcher than the large, faceless corporate entity. If we are at all principled, it's easy to imagine ourselves in a similar situation.
Why the hand-wringing over this email they sent, the suspicion that it was all "hardball negotiating" and "bluffing"? The email to the mailing list pretty clearly outlines their concerns and reaction to Google, it seems a pretty safe assumption that their email to Google management covers much the same information.[0]
This idea that they are bluffing strikes me as insulting: is there any reason to suspect they are bluffing? I don't see that in any of the coverage, by my reading this person felt this back-and-forth over the publication of this paper at the last minute was the last straw, their exasperation and frustration with Google is clear. It seems to me they were entirely willing to resign.
[0]: https://www.platformer.news/p/the-withering-email-that-got-a...
She's even tweeted that she suspects there was a team of people behind the decision to accept her resignation. All those people don't put Google first. The hard truth is that she was probably extremely unpleasant to work with it (this is my reading of the above link and from what I've witnessed on Twitter) and they had just reached their limits. Not Google, but the people that had to work with her (I'm clearly not referring to her subordinates who are all apparently "indebted" to her which is a whole other issue onto itself). She comes across as very entitled and quite frankly, although I hate to say it as it's a bit of a cliche at this point, narcissistic.
These are all based on what I've been able to find online. I've never met her in person. I'm trying very hard to not see her negatively but it's also hard finding evidence that's she's not hostile to those that don't 100% accept and agree with her. I'm definitely open to seeing her in a different light if anyone wants to share links. However, just because you research ethics doesn't mean you practice what you preach nor that you're a saint who can do no wrong.
Would that be somehow bad? I'm not even entirely sure of what does that even mean. If I submit a resignation letter, I don't even expect it to be rejected, engineers and researchers are not some kind of ministers. At most some people can consult each others and ponder if its worth to attempt to retain her, but even then the concept is not entirely one of "I/we don't accept your resignation letter"...
https://imgur.com/a/1P7HQFW
I also saw the tweets at Yann LeCun. I'm not his biggest fan, but that wasn't right. She's hurting minorities for her own agenda.
Edit: removed info about myself.
Unless you still work at Google, how do you know she actually broke any rules? As far as I know, her e-mail to the Brain list is not public?
Yes, it is a bit absurd. And if you're the first minority hire on a team that hires 1 person a year, you will be paid under the team average for a while, just as if you weren't.
> And at what level are we talking about?
Entry level employees at FAANG mostly. For PhDs managing people, such as the person in the article, I expect there's a much wider band of possibilities.
You can absolutely get paid that much more. Companies make exceptions all the time and justify two people doing the same job by paying more if someone is more experienced even if the technical skills performed and responsibilities by both people are equal. Instead of being rewarded for working twice the hours to learn something faster due to whatever reason someone can still be paid more for basically accomplishing the same amount of progress in their career but taking more time to do it.
Trust me there are all kinds of reasons you can get paid three times more.
What does this mean? That you don't feel like the race you actually are because you don't play the race card?
I hope I misread that and there is a more charitable explanation.
By virtue of their research area, her and her team no doubt push against this limit very often. Based on this, I’m not sure it’s possible to really pursue many lines of research inquiry into AI ethics and fairness within a place like Google.
Yes. [1]
> Had she accomplished something of great distinction in this field prior to google?
Yes. [1]
[1] https://scholar.google.ca/scholar?hl=en&as_sdt=0%2C5&q=timni...
This move makes it clear that Google has no commitment to Ethical AI as a field; they want rhetorical cover for their ongoing bad actions.
Such an inspiration!
That's the deeper question ML research/industry has to grapple with. When these models are deployed increasingly quickly and at scale in ways that can potentially cause massive harm, why is it okay to keep doing the same exact harmful things?
LeCun mentioned that there would have been opposite problem if it were trained on a dataset from Senegal. But why wasn't it trained on a dataset from Senegal? Why do we always see these errors where white-centric datasets produce white-centric results?
It's obvious that while it would be a symmetric situation a vacuum, we do not live in a vacuum. We live in a world with deep sociocultural biases in favor and against various racial groups. And it is unjust to let AI perpetuate and entrench these biases by acting at with this bias at scale.
Acknowledging dataset bias by itself doesn't address the bias meaningfully. In practice, we often treat the bias as an exogenous factor when it is not, moving it outside the scope of our responsibility. But it is very much the product of our work, a reflection of our choices, values, and beliefs about what to prioritize. We can't abdicate our responsibility for it (even if we choose not to prioritize it.)
That said, how should a person like Yann LeCun argue his case? Namely that biased models are the result of bad datasets more so than the result of bad algorithms.
I don't think Yann would disagree that biased datasets is a large systemic issue. How should a person like Yann make his point?
It doesn't seem to me that Yann and Timnit disagree all that much. They both agree biased datasets is a problem. They both agree it's a systemic problem. They both agree it does tremendous harm when these biased models are deployed.
I'm very confused as to why there even could be debate between these two people. I cannot find any meaningful difference in their views.
Probably because it was a research AI not a production AI. Having a very diverse dataset at that stage doesn’t help with your research, so it is fine to use whatever is easily available.
At that stage, you are trying to show that your approach can work in some cases. Once you've got that, it is time to expand the research with a wider range of inputs to find out what the limits of your approach are.
For example, if I were trying to make a US English speech to text transcription system, I might start with recordings of assorted NPR programs, because NRP often makes the recording available online along with they transcripts.
That would be great for determining if my basic approach has promise. Once I have determined that, so know that the whole endeavor is not just a waste of time, I could go looking for data that includes speech that has characteristics that would be missing from the NPR data, such as heavy regional accents.
Thanks the comment and this explication, you clarified the conflict that incident for me.
While I think the sibling comment is correct that LeCun and Gebru would agree on the proximate causes and mitigations of the adverse outcomes of that particular super-resolution model, the issue was (seemingly?) that focusing on the proximate causes can be read as absolving researchers of their responsibility for those outcomes, and avoids discussion of that responsibility in any instance.
Which is a entirely fair criticism of the field as a whole, though it may have been a bit lost in translation to the particular avatars in that instance.
I’ve always thought only SV has the freedom and power to start this fight. Then academia and other institutions can follow and ride the momentum.
I'd suggest given Timnit is choosing to only tell a very small part of the story here, and Google is obviously not going to publicly comment on an HR matter, especially one this sensitive, that it's best to withhold judgment.
Great stance to take bruh.
But judging from her tweets this is not happening, bruh.
I quoted the relevant parts here: https://news.ycombinator.com/item?id=25292266
Assuming Dean was aware, this civil discussion should have happened either in the meeting described, or shortly thereafter. So far we only know her story on this meeting, but that story sure makes it sound like there was no discussion to be had.
Hopefully some clarification comes out about what her demands were and the internet drops their pitchfork about something that happened long ago
https://twitter.com/timnitGebru/status/1331757629996109824
Right now the only way to disagree with unethical actions by your employer is leaving; easily replaced.
I'd rather they kept their powder dry to protect employees fired for things like union organisation, and protecting the workforce during large layoffs/reorgs, individuals discriminated against for their race/religion/sexuality, being compelled to take actions that are immoral/illegal etc. None of which seem to apply here.
Who ends that sentence with a smiley?
I respond to stressful and/or tragic situations with sardonic/dark/derisive humor. It's a coping mechanism for me. Some people cry. I crack jokes.
Just yesterday I said in a meeting with my boss 'No one wants to touch that project, because the customer gives out negative feedback like cheap halloween candy.'
If I were fired from my job in a way I felt was unfair, I'd probably do something similar.
https://twitter.com/timnitGebru/status/1334341991795142667
> Apparently my manager’s manager sent an email my direct reports saying she accepted my resignation. I hadn’t resigned—I had asked for simple conditions first and said I would respond when I’m back from vacation. But I guess she decided for me :) that’s the lawyer speak.
> 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. Then she sent an email to my direct reports saying she has accepted my resignation. So that is google for you folks. You saw it happen right here.
In the email she posted, it seems that they "accelerated the timeline" for her termination because "certain aspects of the email you sent last night to non-management employees in the brain group reflect behavior that is inconsistent with the expectations of a Google manager", i.e. a separate email from her conditions email.
It seems like she was ready to leave if they didn't meet her conditions, and that this would have been a reasonable outcome for her. However, it bothers me that they wouldn't even come to the table about the non-condition email.
Like OK, "inconsistent with the expectations of a Google manager". That's some weak shit. It had better be something pretty fucking severe for them to fire her immediately and not, I don't know, do some planning around transitioning her projects? Work out what to do with the team she herself managed? Be given the grace of leaving on a good note that an engineer of her caliber deserves?
She's a leader of their ethical AI group. What happens to that group and all of their research and direction now? Who do they hire to replace her? This isn't the same level of impact as firing a cog in the wheel IC who gets too uppity with organizing.
This stinks like a convenient excuse to shoo an AI ethics researcher asking some inconvenient questions out the door.
https://twitter.com/mjg59/status/1334382463498391557
Neither of us are on her team at Google. (I'm certainly not and I doubt you are.) We have no idea how easy it is to work with her.
Certainly if I were working with an AI ethics researcher I would expect them to ask difficult and uncomfortable questions. I think that's part of the job description.
Not fired because that's a real job and not a diversity hire?
That said, Garrett makes clear his ideological sympathies on Twitter. Undoubtably comments like this one don't endear him to his management chain:
When @computerfemme was fired, the VP of security at Google Cloud sent out an email to the security organisation containing one absolutely false statement and one misleading statement, both of which portrayed her acts in a more negative light.
Pretty serious stuff. He doesn't elaborate on what those claims were. The woman in question was fired from the security team for violating IT security policies: in her words, "All I did was make a popup to share the labor notice Google has to share with its workers", something that Google seems to have recurrent problems with staff doing. In other words she was rewriting the contents of web pages to meet her own political objectives.
If I were the VP of Security at Google I would have serious concerns at that point about insider threats from sympathisers on the security team. You can't be publicly slamming your own VP as dishonest for firing someone who subverted IT policies if you actually work in security.
And yet, somehow, he’s not enough of an “insider threat” to be dismissed.
This paragraph that you just wrote is one of those claims that are wrong and/or misleading
The Google security team is frankly deeply worrying at this point, and I say that as someone who used to work there on security related topics (but consumer account security, not internal). A small number of them have privileged access to mandatory Chrome extensions that are used for various security objectives, yet they seem to have no qualms about abusing them to advance their own political agendas. Moreover they don't recognise that this is a problem.
Controlling Chrome extensions that can inject JS into arbitrary websites equals root at google.com because everything they do is admin'd via web UIs.
What if you're a law firm who advises companies faced with union action? Is it safe for you to use GSuite or whatever it's called today? My guess is no, because the people who ultimately control Google's IT infrastructure are (a) willing to break the most basic of IT security rules and then boast about it publicly, and (b) many of them are politically extreme by the standards of most of the world.
Most astoundingly of all, I don't believe this is the first time Google has fired people for injecting JS into websites using internal security mechanisms to advance political agendas. In any normal company there'd be a mile of controls, processes and guard dogs surrounding any mechanism that could do this by now as it's been abused before. Apparently Google Security do not have their act together on this.
Does not mean rewriting the content of webpages, or having "root at google.com". It's better if you'd stop such speculation.
The reality is much more pedestrian than what you describe: the person that was retaliated against only wrote simple, inert, strings of plain text.
That text was not injected anywhere, and there was no arbitrary code involved.
In fact, everything was safe, and went through the usual security controls.
The only "danger" in having that, is that people can read it.
Ultimately, it's laughable how you think that informing other coworkers about the NLRB is some kind of politically extreme agenda. It's clear that you have an axe to grind
Please do explain how someone makes a message appear in your browser when you visit a particular website, without having access to a highly privileged browser extension? That can at minimum monitor the URLs you're visiting?
I honestly really hope you aren't on the Google security team, because you seem to be missing the issue here. Someone who had a high level of access was fired for violating basic trust, apparently with the collaboration of her coworkers. What other things could this person have done to advance her political goals by subverting her access for things it wasn't meant for? Outsiders can't know. All they can see is that the company doesn't have a grip on its own workforce. That is not confidence inspiring.
Yes, that's crazy... And that's exactly why people have been angry at the retaliation
> Please do explain...
To get more details, the appropriate thing to do is to draft a post that explains how these things are accomplished at Google and get it approved for publication by the legal review team. I've never done it and, while I don't exclude that I might do it, it probably won't happen soon.
Hiring has apparently failed to meet goals with zero consequences suggesting a lack of concern with minorities.
I don't think someone misinterpreting disregard for ethical ai research as disregard for minorities is slanderous, heck it might be charitable as the later is known.
The casualness to make this now about race/gender as well when that wasn't even her inital argument.
I can't imagine working in this kind of environment. It must be exhausting, always having to tread on eggshells or end up on a twitter firing squad. Not being able to argue points, without it ending up being about race/gender.
Do we really have to pretend that someone blessed to be working on AI at google with the greatest minds on something so exciting is disadvantaged compared to the rest of us?
I don't see how people can deny that people take you more seriously on tech stuff if you're a man. It's pretty obvious to me, honestly.
Equally obvious, you're not going to become a top AI researcher without being immensely talented.
It always was (also) about race/gender!
That said, I don't know Timnit, and I don't know her specific circumstances... I believe that Google is full of people who wants to solve these problems, but somehow leadership fails to capitalize on that. It might only be a few bad apples in the leadership, but each one wields a huge amount of power (hiring and firing people)...
These really exist? How do you know when you've reached perfect diversity? I'm honestly curious.
Is it based on population distributions? What would be an example diversity OKR
https://www.blog.google/perspectives/melonie-parker/how-rete...
Many Amlaw 500 firms have agreed to the Mansfield rule[0].
> Now in its third iteration, the Mansfield Rule Certification measures whether law firms have affirmatively considered at least 30 percent women, attorneys of color, LGBTQ+ and lawyers with disabilities for leadership and governance roles, equity partner promotions, formal client pitch opportunities, and senior lateral positions.
Mansfield rule certification is independently audited on a biannual basis.
It's also common for clients to have their own diversity requirements[1].
[0] https://www.diversitylab.com/pilot-projects/mansfield-rule-3...
[1] https://www.diversitylab.com/knowledge-sharing/clients-push-...
The VP that fired her/"accepted resignation" wrote to her reports about it:
https://mobile.twitter.com/alexhanna/status/1334348137616568...
It's frustrating to realize, but unsurprising: the people higher up in the hierarchy will always have more power than those below.
Here's a tweet from Timnit Gebru mentioning that she speaks Tigrinya, presumably natively: https://twitter.com/timnitgebru/status/1078042727936376833?l...
So it appears to be a Tigrinya name. TIL.
(If they're not public then I imagine they'll be leaked eventually if this gets enough media attention. Are we witnessing the opening act of Google's next enormous Damore-style PR clusterfuck? Grab your popcorn.)
I personally have ever seen the 'do x or I quit' move work exactly once. The second time they tried it they took them up on the offer. They were even very nice to work with and were part of the 'do not fire this person or it is chaos' team. When he did it the first time he said 'I can only do this once if I try again they will fire me on the spot'. He did it a second time just to see if they would match his external offer (they let him ride the 2 weeks at least). You usually only get one shot at that sort of move. If you do it all the time, managers start looking for ways to offload you. You will be seen as a threat to org stability as you come off as unreliable. No matter how key you are.
[1] https://twitter.com/negar_rz/status/1334369747241218050 [2] https://twitter.com/dylnbkr/status/1334395186705702913 [3] https://twitter.com/L_badikho/status/1334393782310227970 [4] https://twitter.com/alexhanna/status/1334348137616568321 [5] https://twitter.com/dylnbkr/status/1334372430437994500
Edit: and here come the downvotes :/
AIs were accused of being biased so Google may have hired her as a defense mechanism. Her job was to justify Googles actions, basically a PR job and not a control organ.
Dunno where you've worked but people who are disruptive to team functioning can (and should!) be terminated. Nobody should have to put up with a hostile work environment.
That does not create a hostile work environment. It's healthy and normal. I would expect an ethicist to fall under a similar category.
Have you considered that she has to remain vague due to the potential litigation coming down the road? In fact didn't she leave some tweets hinting at that ? ("Everything I say will be used against me ..." ?)
This seems to be it though, and I quoted some (I think) relevant parts of what she's criticizing:
https://news.ycombinator.com/item?id=25292266
https://www.platformer.news/p/the-withering-email-that-got-a...
(2) I don't understand how someone can make public accusations without providing the full picture. Maybe she is legally constrained, but then she shouldn't have said anything and handle the whole situation legally first and then write what she wants to write about it. Google might be wrong here, but we definitely cannot see that.
P.S. It seems to me that the only person who handled "firing" professionally was the most ridiculously dressed person on the planet who goes by the pseudonym Dr. Disrespect. So much drama going on around nowadays.
However, the main problem with Timnit and her colleagues is that they come into every situation assuming racism even when the situation is ambiguous. Likewise, if you disagree with her or want to have a nuanced discussion, she'll accuse you of imposing intellectual labor on her and in some sense you are racist for forcing her to engage in discussion.
Again, so while I appreciate how she is purportedly working to make my life better in tech, there needs to be a real awakening as far as giving people the benefit of the doubt and allowing for discussion without shaming. Google has accumulated many people with this type of attitude who have created a culture of intimidation. You can see all of this happening in real time as her colleagues come to her defense knowing little about the situation and assuming that she was fired because she was black.