Read on Twitter
I’ve picked up a bunch of new followers in the past few days, and I suspect many of you are here because you’re interested in what I might have to say about Google and Dr. @TimnitGebru. So, here’s what I have to say:
Dr. @TimnitGebru is a truly inspiring scholar and leader. Working with her these past few months has been an absolute highlight for me: https://twitter.com/emilymbender/status/1334875708355530753?s=20
I was super excited to work with her because of all the excellent work she has done. For highlights, see this thread compiled by @math_rachel: https://twitter.com/math_rachel/status/1334545393057599488?s=20
I expected it to be awesome working with Timnit, and I was right!
The paper started as a conversation between the two of us but expanded to include a larger group, representing different research traditions, to provide a multi-viewpoint investigation of the possible risks of large language models. https://twitter.com/_KarenHao/status/1335055489168797697?s=20
Google used the paper as a catalyst, but it’s hard to see how it’s really about the paper. (If their goal was to squash this paper or at least not associate it with Google, well…)
It’s also been distressing, to say the least, to hear Google claiming that she resigned. Here’s @GoogleWalkout setting the record straight on that point: https://googlewalkout.medium.com/setting-the-record-straight-isupporttimnit-believeblackwomen-5d7bbfe4ed90
And a thread from @rajiinio gathering it all together, in case anyone needs further evidence that she didn’t resign: https://twitter.com/rajiinio/status/1335410726656237568?s=20
Questions I’ve been asked a lot in the past few days include: What does this all mean? Where does it end up?
The most important story here is about the barriers put in front of Black women in tech (and in other sectors too): https://twitter.com/reddyourmind/status/1335994105529307137?s=20
A best-case scenario outcome then is that this story, Timnit’s bravery and clarity in speaking out, and the attention it is getting will serve as a catalyst.
To the vast majority of my followers who aren’t Black women: The next time a Black woman at your workplace tells you something you don’t want to hear, are you ready to listen?
A secondary story here (and the one I’m most often asked about) is what Google’s attempt to squash the paper means for AI research and especially AI ethics research, given that so much of it takes place in industry.
My take is that moving to a future where the tech we build addresses and uproots systems of oppression (instead of replicating or further entrenching them) is an all-hands-on-deck task:
We need researchers like Dr. Gebru and the team she has built at Google working in an environment that gives them visibility into the corporate processes, but with the freedom to turn a critical lens wherever it is needed.
We need people engaged in product development who both have the skills to envision possible adverse impacts and are empowered by their corporations to bring that input to development teams. See @RadicalAIPod episode w/ @baxterkb for a great exploration: https://twitter.com/emilymbender/status/1335384009866506240
We need people in the academy engaged in this research free from corporate incentives but also training the next generation of technologists to understand their work in its social context.
And finally, we need people who know how to design effective regulation to be informed by those who understand how the tech works and how it impacts people AND people educating the general public on what to advocate for.
If brilliant scholars like Dr. @timnitGebru have to put all their energy into fighting workplace discrimination, it’s going to take so much longer to get to where we need to be.
Google using our paper as an excuse to fire Dr. Gebru is a set-back, for sure: at the very least it inhibits her team’s ability to pursue their excellent work.
In addition, it inhibits others’ ability to learn from and build on what they do, as @_alialkhatib articulates here: https://ali-alkhatib.com/blog/on-google-and-intellectual-freedom
The best-case outcome I can imagine on this point is Google taking concrete steps to support the scholars on Dr. Gebru’s team and setting up MUCH clearer processes around the approval of publications, which make clear that it’s only about checking for disclosure of IP.
But I don’t see how they can possibly to that without owning the fact that it wasn’t the paper, admitting that she didn’t resign, and coming clean on the actual reason they fired her. Which, also, would be part of my best-case outcome.
