At that point she was on her way out, one way or another; most employers would rather not work with people like this.
Having said that I still feel Gebru is extra-ordinarily self-entitled person. Few months ago she insisted that Jeff Dean take her side in Twitter feud she initiated. She has the mindset of either you are with me all the time or you are racist+misogynist. She complains about "micro agressions" towards her from everyone but she herself is super aggressive. When disagreement arise, she marks you down as enemy and launches attacks to destroy you completely with sole purpose of making you homeless, jobless and making sure your kids die of hunger. I wouldn't wish upon her as colleague to anyone. You might have right to disagree with me but this is what I have felt. I still take her side in this current episode nonetheless because I think she is on the right side.
Ultimately we are all playing opinion game here and only few people knows what really happened.
She can try to sue them, but now that she made it public it's quite unlikely any court would look at it.
All in all, it's obvious Google didn't follow their own supposed research protocols, but this whole thing is much more complex and I think it has very little to do with her paper.
She still had a job after she published the paper, she still had a job after she told people to stop working, she stopped having a job after she demanded confidental HR records of people who feared giving her rather simple feedback.
Sample comment:
Another ex-colleague here. I was not going to participate in the discussions but your post made me realize objective truth should come out. I do believe she actually thinks she is making the world a better place but in reality any interaction with her has been incredibly stressful having to carefully weigh every move made in her presence. When this blows over her departure will be a net positive for the morale of the company.
To give a concrete example of what it is like to work with her I will describe something that has not come to light until now. When GPT-3 came out a discussion thread was started in the brain papers group. Timnit was one of the first to respond with some of her thoughts. Almost immediately a very high profile figure has also also responded with his thoughts. He is not Lecun or Dean but he is close. What followed for the rest of the thread was Timnit blasting privileged white men for ignoring the voice of a black woman. Nevermind that it was painfully clear they were writing their responses at the same time. Message after message she would blast both the high profile figure and anyone who so much as implied it could have been a misunderstanding. In the end everyone just bent over backwards apologizing to her and the thread was abandoned along with the whole brain papers group which was relatively active up to that point. She has effectively robbed thousands of colleagues of insights into their seniors thought process just because she didn't immediately get attention.
The thread is still up there so any googler can see it for themselves and verify I am telling the truth.
You don't hire an ethicist to make you look bad. You hire them to make you look good at what you do: diversity in AI. Gebru failed at her job.
But that would also mean Google exploited her for that.
I can't understand WHY so many people obviously think that it is totally fine to be behave utterly unethical, especially when a person whose behaviour is in question is "professional ethics researcher". Looks like they think that a big goal justifies the means. Welcome to Russia circa 1917, or Germany of 193x.
Insanity((
You admit she's a bully, super aggressive, constantly makes spurious accusations of racism, attempts to destroy her enemies totally and is in general a nightmare employee in every way. Then you say you take her side because Dean could have given her "transparency"? Are you aware she was demanding the identities of people who had criticised her work, presumably so she could then go and bully them too?
This person is a 100% A+ straightforward case of someone who should be fired and in fact should have been fired long ago. All those attributes are the sort of thing that absolutely qualify someone for dismissal. If you were feeling trollish, you might even say it's black and white.
Given her role is likely just for PR it seems rather foolish not to at least give her two weeks notice or something.
I mean, I don't see how google could NOT terminate her after such an event. I am in a nation with extremly strong worker protection law but that action alone would be enough to fire you on the spot.
They are not as strong as you think then.
In countries where there are strong work protections law striking is a right, you don't get paid for the time you don't show up at work but you can obviously convince an entire department or an entire company workforce to go on strike, without consequences.
Without it the protection is not strong, it means an entire company against the individual worker, which is obviously unfair (regardless of what you think about work protections law, Google against a single human being it's an unfair battle, that's why "strong protection" means that the worker deson't have to fight alone and can call other workers to join the fight).
If I wrote an email like that not only my union would not protect me, they would most probably tell me the employer took the right decision because I put at risk other employees positions (if they followed my reccomandations to stop working just because I said it)
This not some crazy conspiracy that she got fired. It would be noteworthy if someone who did what she did didn't get fired.
Google lured many of these researchers away from academic roles, with the tacit promise that they would be able to continue their research - and at times political - endeavors relatively unencumbered. Google has built a research empire on this promise, which has been responsible in part for their continued dominance in various markets.
My point here is that these researchers have a lot of leverage. Advocating for action within an organization, counter to the current immediate goals of management, is only taboo (and thus “fireable”) in most companies because workers have almost no leverage, generally speaking. Not because it’s some inherently “bad” thing. If Google becomes seen as a place that fires researchers for advocating for their beliefs, especially in a capacity that they’ve been explicitly hired to do, then I think a lot of researchers would - and rightly should - reconsider whether such an environment is consistent with their values as academics and, often times, activists. Google’s research group is enormously valuable to the company, and there’s only so much reputational damage of this nature the organization can sustain before academics decide to take their talents elsewhere.
Serious question: what makes them strong if employers can't organize a simple strike?
You have to follow strict procedures, work with your union and go into one or more formal negotiating tables with all the representatives of all the interested parts before even thinking about giving an ultimatum like that. And for sure you cannot communicate in that way using directly your work email
Because in my country you can't be fired for using the work email for that.
Communicating with other workers using their work email is absolutely allowed, courts ruled about it several times [1].
I usually receive trade unions (there are several of them) communications on my work email from the unions' work email, because they have been authorized by the company to send them.
In this case she wrote directly to workers using work email, but having no union or labor protection laws it doesn't make any difference whether she could or couldn't, she could be fired anyway without having to provide any reason.
In countries where there are strong laws protecting workers she would have written to the union members and they would have done the same thing: write directly to the workers.
Of course she did it at Google so it is different, but here the strict procedures to call a strike are only necessary if a public service that requires continuity risk to be interrupted, otherwise unions are only required to alert the company that the strike is going to happen, but have no requirement whatsoever on how to organise it.
Which sound logical to me,strong protection means IMO freedom to collectively counter the actions of the company, if that's not allowed the protection is not strong.
[1] Court of Catania, Labor Section, February 2, 2009 "The RSU employee can send trade union communications by e-mail to the employees of the company during their working hours and to their company e-mail address using his personal e-mail address"
Perhaps.
But doesn't explain stagnation in Japanese economy where there are no Italian unions.
Or why France did much better despite having even stronger work protections laws than Italy and wilder strikes (the gilet jaunes for example) or those happened last year against the pension reform where the workers of public transportation went on strike - for weeks - without even announcing it.
Even countries like Germany, Singapore, Switzerland and Finland are doing worse than Pakistan in the GDP growth race.
My point was than if there are stronger work protection laws somewhere else and the laws of your country are weaker, they are not very strong, they are moderately strong.
Japan seems to have evolved a work culture very similar to strongly unionised societies but without unions. Japanese salaryman culture is famously a culture of employment for life with unusually strong loyalty between employee and employer, hence weird things like "banishment rooms" that you don't find elsewhere. If it's the end results that matter and not the means, Japan might not be a good counter example.
I think French strike law sounds tighter than Italian strike law? The French are famous for striking but strikes must still be a collective decision and related to a specific set of issues, whereas Italian strike law really does sound incredibly broad and vague.
With respect to Pakistan, that's doesn't mean anything, poor countries always have very high GDP growth. It's easy to grow something small and backwards by a lot because you can get a lot of relative growth just by copying what other countries do, and less absolute improvement is needed to get a percentage point of growth to begin with. You can only compare GDP growth rates between countries of a similar level of wealth.
Of course, if you leave on good terms, your superiors might look at you kindly and work with you to meet not just their needs, but also your needs. That's why it pays to be humble and generous in your interactions with other people.
You can as an employer walk away whenever you want, you just don't get the salary for the difference between the notice and when you left.
Example: if you have to give 30 days notice (very common in Italy) and leave on the spot, after your resignation have been accepted, you have to give the company the equivalent of 30 days of pay.
Which is exactly the money they should give you aniway if you stayed for the entire notice time.
But usually the employer and company find an agreement before coming to that.
Google exercises a subjective standard to determine whether or not an employee is using mailing lists inappropriately (e.g. James Damore).
Source: ex-google
You realize that this is just an argument that whoever is in a position of authority over you is right by virtue of having that authority?
Is insubordination to represent your responsibility area, as Ethical AI Researcher?
Is insubordination to send an email in order to establish dialogue around matters of disagreement between professionals?
After reading, it seems very much in line with her role as an ethicist. She merely states that there are various risks associated with processing large datasets.
Its a bit sad that they attempted to censure such an innocuous paper that without malice states “there are risks to consider.”
What's the strongest argument against her "merely" doing this?
As I've stated in other post, this is very, very troublesome for AI and Google.
OTOH, Google has much emphasis on ethics in AI/ML otherwise. But the whole company, especially management, need to put actions in alignment with their words on this matter.
Such behaviour, unless corrected, undermines Google's entire AI/ML and cloud operations.
It's called steel manning.
However, without revealing names the paper outlines internal company matters, presenting opinions and tautology that do not fit proper format of professionally published research. Instead of reworking a flawed paper, the researcher escalated and exposed individual opinion, publicly. Such company representation is out of scope for research positions. As Google could not agree to terms of email, it was taken as resignation notice.
I believe we should have no curruption in Russia. By your logic, unless I can give a strong argument in favour of corruption, I am not prepared or not honest?
Are you prepared to give a rigorous argument in favour of: terrorist attacks, slavery, eugenics, facism, child labour, sexism
> If you can't argue the counter of your own position in the most honest way possible then either you're not prepared or not honest.
I didn't say that if your position is on the opposite side of child abuse and you don't want to change your position that you aren't honest.
Yes, I'm prepared to try and make the strongest argument possible for things I may not fully understand, until I do. Whatever those things make me feel.
Sometimes the strongest arguments have large flaws (mostly moral with you examples) but sometimes we just don't understand the opposing arguments and then just create strawmen to rage against.
I'm opposed to this type of irrational emotional behaviour. It causes things like mob violence which I'm very familiar with.
So when someone can't articulate the opposing views then they need to be warned or directed to do so. If they refuse and run back to a strawmen then yes they are not honest and obviously not prepared.
uMkhonto we Sizwe was a terrorist organization. But they had a very strong argument for why terrorism was an appropriate action.
I don't say that I agree with their use of violence, but they had an argument that makes blanket statements like "terrorism is evil" harder to make.
https://analyticsindiamag.com/how-is-ethical-ai-different-fr...
https://www.kdnuggets.com/2016/03/ethics-machine-learning-ta...
The strongest argument for her not doing it is keeping her job vs maintaining her professional responsibility as an ethicist, no?
I don’t understand how we arrived at some conception of workplaces as little dictatorships, where bosses can treat people however they want and that’s somehow good or legitimate.
If you want out of the contract the boss cannot make you do the work. You have agency, they cannot threaten you with violence as a dictator could.
In my opinion, might doesn’t make right. Managers aren’t de facto right purely by virtue of being a manager.
When it comes to who has the ability to steer subordinates, managers are always right. Except if their managers disagree...
They have complete authority... They don't have to have someone else endow it to them or give it to them...
So yes, they are by definition authoritarian. That's one of the cornerstones of being able to own private property.
But my point is that all of your arguments rest on an unquestioned assumption about how ownership translates to authoritarian control of others without explaining how or by what justification that power is arrived at.
The answer of course is that it’s a social construct enforced by states to varying degrees, present and historically, and thus no way inherent to the thing itself. Appeals to status quo are therefore not justifications for the validity of the relationship, as it has changed over time, proving malleable to all kinds of political and social forces (e.g. the abolition of slavery, the formation of unions, worker collectives). Defending an authoritarian relation to property as it relates to other people’s labor thus require an actual and explicit defense, which you have so far not provided.
Very much so for refusing to do work assigned to you.
And it doesn't mean "bosses can treat people however they want". It means "you are employed to do a job, if you refuse to do that job, you are no longer employed"
Also, "do my bidding" should be written as "do your job."
I'm from a country with strong labor protections (striking, 3 strike dismissal etc.), and you're moving the goalposts.
The UK is an example of a European country with relatively tough union laws. The list of issues that it's legitimate to strike over is restricted, votes are required, the general secretary of the union must authorise the strike and so on. Strikes do happen of course but not because some random middle manager threw a hissy fit about feedback on a paper she wrote.
Gonna take a quick guess you're still in undergrad.
Yes, this is literally how employment works. Your future boss's job will be to use your labor efficiently for the company. You may disagree with your boss on an assignment, in which case it would be wise to schedule some time and professionally discuss your concerns. Your boss may be persuaded, and good bosses will try their best, within reason, to keep their subordinates content with the work they're doing.
However, at the end of the day, the boss may disagree with your points and still require you to do the job as assigned. If you refuse, you will likely be fired. Note that you also have agency here: If you don't like the work the boss makes you do, and you feel like your concerns are consistently falling on deaf ears, you can quit.
What is your proposal?
Her clout/skills/PR and her attitude are a package deal.
Google wants to use her for PR, but expects her to be a lapdog.
I conflicted about many things, but respect her for having backbone.
"Do these things or I'm gone and taking my team with me"
and
"These things are so fundamental to my role as an ethical AI researcher that it's difficult to see myself doing this job without them"
IANAL, but I suspect a court would also interpret them differently.
And even so, Google likes to position itself as a nice employer, which listens to their employees' concerns. De-escalating the situation would have been nice - immediately terminating the concerned employee makes Google seem like another giant faceless emotionless multinational corporation (which it may well be).