Timnit Gebru on her sacking by Google, AI’s dangers and big tech’s biases

> The thing that was very confusing to me as an immigrant was that liberal type of racism,” she says. “People who sound like they really care about you, but they’d be like: ‘Don’t you think it’s going to be hard for you?’

This is why I object to teaching kids about racism in schools. Maybe the teacher was racist. Or maybe she got good grades—which isn’t hard to do in American high schools—but her teachers didn’t think she was one of the exceptional kids who could go on to excel in the field as a career. If you aren’t socialized to think of yourself as the target of racism, the natural response is to probe further. “But my grades are good!” The explanation that comes next might lead to learning and growth, or at least renewed motivation (“I’ll show him!”) But if a kid has been socialized to perceive such interactions in terms of racism, there are, they can’t learn from such conversations. After all it’s pointless arguing with a racist, nor can you learn anything valuable about yourself from a racist.

I moved to Virginia from Bangladesh back in the 1980s, to a school and neighborhood that was nearly 100% white. At the time, northern Virginia where we were was solidly Republican. But my parents never taught me about “racism” and it was a tremendous blessing. I never wondered whether the ordinary negative interactions of life were racially motivated. Was that kid being a bully because I was brown, or because I was a book worm? What good would it have done me to recognize it anyway?

I’m extremely worried about how the discourse is going to affect my kids. It’s impossible to protect your kids these days from folks like Gebru.

I find it fascinating that in the middle of a hype cycle here, where really central people to AI are saying we should be worried (Altman, Hinton etc.), there's still just outright animosity towards people who are actually interested in studying these problems. It's also very far from a theoretical problem, did you know that crash test dummies are men? Literally, until extremely recently crash test dummies were all designed to be 50th percentile men[1]. Surely they test women dummies too? Oh yeah! They put a small woman dummy in the passenger seat. But still, they've presumably measured women and designed something appropriate? Nope! The small female dummy is defined as a scaled version of the male dummy. It's genuinely astonishing how short sighted engineers can be in designing technology and it's wild how angry they get when anyone tries to actually fix it.

[1]: https://www.sciencedirect.com/science/article/pii/S000145751...

"AI threatens to deepen the dominance of a way of thinking that is white, male,"

I stopped reading there.

I'm white I'm male I have 0 advantages because of that. Always the same lies everywhere. I'm tired of it.

> After her departure, Gebru founded Dair, the Distributed AI Research Institute, to which she now devotes her working time. “We have people in the US and the EU, and in Africa,” she says. “We have social scientists, computer scientists, engineers, refugee advocates, labour organisers, activists … it’s a mix of people.”

Virtue signal by donating to my slush fund.

Large models are indeed dangerous, because they threaten to concentrate power in the hands of a small number of tech companies with the resources to do the training, and because, once those models exist, they devalue human labor.

Some technologies are leveling. It may be debatable, but I've heard it argued that the musket was one of these. The musket was cheap. It rendered expensive mounted plate-armored aristocratic warriors obsolete. Down came the whole feudal hierarchy necessary to keep those horses fed.

These AI models are the opposite. They do not spread power around; they centralize it. They increase dependence.

So in many ways she's right.

What I would like to ask, though, is -- instead of deepening the intersectional cuts that were used to dismember Occupy, what can a technologist do to both reduce centralization and increase solidarity? Is it possible to have a broad-based populism that isn't particularist? What technologies could be in its arsenal?

Possibly AI (which is really compression) can be part of this if it enables offline, disconnected "compressed Internet"s to exist off-the-grid, like solar panels?

How do we reduce dependence on the platforms? The different ethnic factions wouldn't fight each other if they were more self-sufficient. If they weren't made to compete with each other for access to a small number of economic chokepoints. How do we destroy the chokepoints? "Fuck your Suez, your Panama -- we have airplanes."

Here is Jeff Dean's email about the situation when it happened

https://www.platformer.news/p/the-withering-email-that-got-a...

Hi everyone,

I’m sure many of you have seen that Timnit Gebru is no longer working at Google. This is a difficult moment, especially given the important research topics she was involved in, and how deeply we care about responsible AI research as an org and as a company.

Because there’s been a lot of speculation and misunderstanding on social media, I wanted to share more context about how this came to pass, and assure you we’re here to support you as you continue the research you’re all engaged in.

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.

Timnit responded with an email requiring that a number of conditions be met in order for her to continue working at Google, including 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. Timnit wrote that if we didn’t meet these demands, she would leave Google and work on an end date. We accept and respect her decision to resign from Google.

Given Timnit's role as a respected researcher and a manager in our Ethical AI team, I feel badly that Timnit has gotten to a place where she feels this way about the work we’re doing. I also feel badly that hundreds of you received an email just this week from Timnit telling you to stop work on critical DEI programs. Please don’t. I understand the frustration about the pace of progress, but we have important work ahead and we need to keep at it.

I know we all genuinely share Timnit’s passion to make AI more equitable and inclusive. No doubt, wherever she goes after Google, she’ll do great work and I look forward to reading her papers and seeing what she accomplishes.

Thank you for reading and for all the important work you continue to do.

-Jeff

> The Ethiopian-born computer scientist lost her job after pointing out the inequalities built into AI.

I am glad they included that at the start of the article so I know to completely skip reading it!

Ethics employees are the lowest on the Bullshit jobs scale.

They bike shed all day. Just regurgitate old stories and what ifs.

They never do anything quantifiable, they don't have the technical skills.

Unlike middle Bullshit jobs that do nothing, they instead drag down employees around them.

Case in point here and as we can see Google is currently losing the AI battle.