Again, I'm not really trying to defend Google one way or the other. At the end of the day it seemed like Gebru published a paper that was quite critical of how Google intends to make a sizable portion of their revenue, and anyone should understand that when you do that in a large company, it's going to get extra scrutiny. And it could very well be that the whole purpose of Google's "ethical AI" team in the first place is to just provide a "veneer of morality" over what are some fairly unnerving practices and consequences of AI. But regardless, the honorable thing then for anyone on that team is to simply resign, say that "I don't agree with Google's values", and find someplace else to do your work (it's not like these people couldn't get highly paid jobs elsewhere). Instead they're throwing tantrums.

So basically Gebru was fired for turning Googles fake ethics group into a real ethics group?

She wasn’t fired for trying to publish the paper.

She threatened to resign if her demands were met, so Google pre-empted it by accepting her resignation.

The narrative around her being “fired” is completely false and is a lie that needs to be stopped in its track.

So do you honestly believe that if Gebru were to interview at another company and they performed a background check asking why her last position was terminated if she put "voluntary" it would come back with no problems?

Do you likewise honestly think that if she applied for unemployment services in the US should we be denied because of willful termination from her last job?

Everyone I know who has been fired, when it's reported to the company, the phrasing is always "Person X and the company have come to a mutual agreement", but in my experience it's never really mutual.

I've known many coworkers that threaten to quit over issues, but it still depends on them making that choice formally. Otherwise they are fired.

Well, its like HR. The Ethics group is there to shield the company, not the customer. Apparently she didn't understand that.

It's fascinating how nobody can say how this paper was such an ethical stab in the heart for Google, it's all about process.

The paper's out there! You can read it!

Maybe we live in completely different worldviews, but an academic on an Ethical AI team’s primary obligation should be academic honest and integrity.

If the truth damages the company, then that is the company’s fault, not the academic’s fault.

Google claims to be a place where academics can do research. Censoring a critical paper goes against that. I’m not a fan of Timnit, but in this case she is right to make a stink. It’s not like her job was to write ad copy.

Is "having standards for what Google pursues for publications" really censorship? I think we'd need additional evidence to say so. If this isn't the first paper Google has stopped when an internal anonymous reviewer gives the red light, it sounds like they're just following their own process for quality control. Maybe the debate should be whether it is the best process for quality control.

Wouldn't the honourable thing be to first try to change how Google behaves and only if unsuccessful resign? Why is the assumption that employees shouldn't even try to influence the direction of their company? Why is doing so considered a tantrum?

There is a huge difference between skillfully navigating the internal politics and procedures of a company vs issuing demands based on flimsy evidence/feelings and threatening to take your toys and go home if you don’t get your way (and then throwing a social media tantrum when they call your bluff and accept your resignation).

Sure, and maybe they misjudged their power and overplayed their hand, but why is it that the law is the only guidance for management while employees are held up to a higher standard? If management can do anything as long as it's legal, so can the workers.

A corporation is not obligated to keep you employed nor should they be outside of a legal contract. I’d prefer it stays that way because I don’t want to be compelled to work for anyone or have them be compelled to hire me. They’re two sides of the same coin.

They are two sides of the same coin, but the law in some countries recognises that the employer has more power, and so favours the worker and makes it harder for the employer to terminate an employee without making it harder for the employee to quit.

Making it harder to fire someone makes it harder to hire them in the first place. This hurts everyone because it reduces the number of available positions by increasing the cost of hiring. Give too much power to the employees and you wind up with corrupt unions that effectively have a cartel over entire areas of the economy. It’s just layers of red tape and misery for everyone. Getting fired isn’ta problem if you can easily get a new job. It becomes a problem when nobody wants to hire because they can never get rid of you or they have to deal with a crooked union. Having high barriers to firing people also makes the fact that you got fired a giant stain on your resume instead of a fairly normal occurrence that could be due to a no fault reason.

Such systems are sensitive to multiple variable, and it's not true that all values except from the current ones lead to a worse result, or, at least, we have no reason to believe that's true. What we do know is that over the past several decades the US, and other Western countries, and especially in the past two, have conducted a massive redistribution of wealth away from workers.

Huh, last time it was contractor treatment issue. Make a show of protest and then get back to desk at work. I wish these world leading AI ethics researchers show some ethical spine and at least go on strike if not resign until management come to talking table and meet their partial and full demands.

> first try to change how Google behaves

By, say, attempting to publish papers and have internal discussions about it?

I don't have all the facts here and neither, I believe, do you, and we don't know what recourses have been tried. My point is just that I find strange the view that the company is entitled to do anything as long as it's legal while employees aren't, and are expected to either leave or stay and shut up.

I think we're entirely in agreement on this. I would say that in all jobs, it's more honorable to first try to change the company rather than quit silently, but if your job description encompasses "ethics" or "compliance", it's pretty much a dereliction of duty to quit without even trying to change policies one finds wrong.

The "honorable thing to do" would be to let them window-shop until they find someone willing to rubberstamp whatever they want?

Didn't they push back on the publication because the paper was lacking in certain regards? IIRC it failed to take into account relevant recent papers, for example. Which probably could have been fixed if they didn't submit as late as they did.

The paper was submitted on time, other papers at Google were submitted for review after they were released. By Jeff Dean's own admission the paper was approved for release.

The demand for retraction was initially given while refusing to state what the reason(s) for the demand were. Only after the meeting where they demanded she retract the paper did they belatedly agree to read her the feedback that they stated as the reason. They wouldn't even let her read it first hand.

Ultimately it seems like Google could have plainly stated their issues with the paper from the start, and that said issues were not so serious that they merited a demand for retraction.

There are a number of things that just don't add up. Even if you take it at face value that Gebru's ultimatum demands were unreasonable, there are several problems with how the situation got to that point in the first place.

Excerpt from Jeff Dean's email:

> Timnit co-authored a paper with four fellow Googlers as well as some external collaborators that needed to go through our review process (as is the case with all externally submitted papers). We’ve approved dozens of papers that Timnit and/or the other Googlers have authored and then published, but as you know, papers often require changes during the internal review process (or are even deemed unsuitable for submission). Unfortunately, this particular paper was only shared with a day’s notice before its deadline — we require two weeks for this sort of review — and then instead of awaiting reviewer feedback, it was approved for submission and submitted.

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

I'm curious: why is ignoring too much relevant research not a good reason to demand retraction? Isn't this one of the most common embarrassing mistakes that inexperienced researchers tend to make? I can fully understand if Google does not want to associate itself with work that might be viewed by the wider research community as shoddy.

> I'm curious: why is ignoring too much relevant research not a good reason to demand retraction? Isn't this one of the most common embarrassing mistakes that inexperienced researchers tend to make? I can fully understand if Google does not want to associate itself with work that might be viewed by the wider research community as shoddy.

For starters if the academic content of the paper were shoddy, that is what peer review is for. Secondly, if it was really about the paper lacking some new context, why did Google demand retraction while refusing to share why? That's what doesn't make sense.

It seems like there's some missing context in the situation because on the surface Google's actions don't line up with their stated reasoning.

Notice how in Jeff Dean's email he made no reference to the fact that they refused to share the reasons for the retraction with Gebru in that first meeting. He also skipped over where Gebru asked for the reasoning and a manager read it to her rather than letting her read for herself.

I just don't understand the secrecy and cageyness if the only problem with the paper was some academic omissions.

That claim by Jeff Dean has been more or less debunked. The real reason that Google demanded the paper not be published is that it is critical of models underlying the company’s fundamental business - the whole message from Jeff Dean is very misleading and states a bunch of other, unrelated things as the “real reason”.

It would be very helpful to point to sources for the claim that Jeff Dean's explanation is debunked, rather than just claiming it has been.

The timing of the submission for review, and the deadline for submission to publication seems like a simple factual statement that should be easily demonstrated as true or false.

Several Google employees said it wasn't the regular process.[1][2]

[1] https://news.ycombinator.com/item?id=25309488

[2] https://news.ycombinator.com/item?id=25309379

AI is far more energy efficient than creating the incalcuable number of human programmers required to directly program the same solutions. Writing NLP code with non-ML methods, to the same level of quality, would be essentially impossible without an operation on the scale of warfare.

Even with the efficiency benefits that it already has, we all know ML still stands to be further optimized, and a lot of people are working on it.

Comparing the exponential increase of "intelligence" of computers versus our own, computers will be as smart as humans with similar energy efficiency in around 50 years, and they require no cars, food, etc. It will be hard to criticize the efficiency of ML when it is already moreso than our biological selves, but it will be done, and we will be wholly outsmarted.

Debunked by whom?

I'm not sure what did or did happen in this case, but I've reviewed enough papers and seen enough of my own reviewed that I know "failed to take into account relevant recent papers" is one of many ways a reviewer who wants to shoot a paper down for BS reasons can use.

To be clear, reading that reason in a negative review can mean at least two things, and I'm not judging which of the two happened here:

1. The paper did fail to take into account the results in some highly relevant and widely cited papers. (I've rejected papers claiming new results myself before now with "this problem was in fact already solved by X, published in Y".)

2. The paper was fine, but it didn't cite every paper published recently on the topic ever, and someone used that to shoot it down for whatever BS reason. Matt Might calls these reviewers snipers: http://matt.might.net/articles/peer-fortress/

According to current and former Google employees on various machine learning forums, Google does not do an academic review on their employee’s papers. That is the job of the journal reviewers. Google usually just checks whether any proprietary IP has accidentally made it in. For them to claim anything different is comically dishonest.

Not true, see christian's tweet. It's also obvious that a paper like this one will receive more scrutiny than a technical one.

Even if the internal reviews aren't anonymous, if you send your manager a letter demanding something or you're leaving, they're probably not going to give you it regardless.

What was the topic of the paper?