Judging from the comments from some of her ex-colleagues, she was not exactly an easy person to deal with:
> The GPT-3 thread you describe was my first exposure to Timnit. Watching that thread unfold left me feeling upset, frustrated, and disappointed.
> I was so excited in anticipation of other Googler's reactions and insights about GPT-3, but that thread got immediately derailed by Timnit into claims of racism, not being listened to, dehumanization, that the whole forum became icy and dead after that.
I'm still trying to gather more insight into what actually happened here. But you have to know upper management calculated there'd be blowback for firing someone of her stature, position, background, race, gender, etc. They did it anyway. That to me says that keeping her around was worse than the storm they knew they'd generate by firing her.
More like, you support the decision, but your immediate boss doesn't, so they start looking for reasons to get rid of you that don't have any official connection to your stance. I remember a while back seeing leaked documents from Google where some managers shared an unofficial blacklist of employees to never promote and try to get rid of because of their political views.
Why are we conjuring hypothetical situations in which you are oppressed, instead of condemning the concrete case where Jeff Dean looked for flimsy reasons to get rid of Timnit Gebru?
It does though. For instance, were any articles framed as [More than X Google workers condemn firing of James Damore], despite that actually being the case? The media, universities and other elite institutions always support the "SJW cultural domination" stance on all these issues and the discussion is always framed from that perspective.
You'll find similar framings such as [More than X GitHub workers condemn GitHub's deals with ICE]. You'll never find [More than X GitHub workers condemn removal of meritocracy rug], despite that also being the case. You may not like me pointing this out, but that's how it happens. It's not cognitive dissonance to see that that's how these things go.
People are afraid of being labeled a bigot and having their careers ruined because they have a nuanced opinion about an issue related to identity politics.
The cognitive dissonance is that a huge number of highly respected people in the field, and coworkers, said "yeah, this was racist bullshit", and the actual paper in question has been seen by a lot of people by now.
But if you are committed to the stance that racism doesn't exist and this is all a grand SJW conspiracy then you have to grasp at straws to ignore the overwhelming number of experts disagreeing with you.
I don't believe in experts. They have no power to change my mind about any social issue whatsoever. So them saying "this is racist" means nothing to me. There is no cognitive dissonance. You're projecting it into me because you think I think like you, that I care about what experts have to say, when I don't.
It was a technical question about the predictive power of language models, the biases encoded in them and the results of their use in a large system like all of Google's connected products.
But I guess to your particular faith-based ideology the mention of race makes it a "social issue"? Anyway I'm sorry I misclassified you and in the process insulted your religious beliefs. That wasn't my intention.
Thank you for your apology. My religion is very important to me and it really bothers me when people think I care about experts. Universities and other buildings of high density intellectual activity should be thoroughly demolished with tanks & tractors (and without the people in them, of course), the charade has gone on for too long.
Gebru and her supporters say that she was fired in part because she's a black woman and the racist Google culture rejected her. If they don't want others to frame it as a "political correctness gone mad" thing, they should drop the race baiting.
