For example, if you start a “women & allies” email list and then fire a Black woman for being honest on it, it probably would have been better not to have the email list in the first place 3/
While teaching girls to code is nice, it doesn’t address the diversity problem we have now: 40% of women working in tech leave, often due to mistreatment, and 50% of Black & Latino grads with CS degrees can’t get hired at all 4/
Sadly, many folks love the idea of little girls coding, but don’t want to work with the adult women software engineers already at their company. There are similar dynamics around race. 5/
So what works? The #1 thing to do is to make sure that the Black women at your company actually have a fair shot at success. This also means you will have to get rid of the people who are actively sabotaging them 6/
Without completing this first step of treating current employees well, even if you are able to hire more women and people of color, you won’t be able to retain them, and a bad reputation would make it harder for your company to attract talent in the future. 7/
When men are mentored, they receive public endorsement of their authority. When women are mentored, they are given advice on how they should change. Guess which of those is actually useful for getting promoted? 8/
Action item: think of 1 person from an underrepresented group at your workplace, and champion their work: advocate for their ideas, praise them to leadership at your company, or recommend them for an interesting, high-impact work project. 9/
Train the managers at your org: don't underestimate the damage that an untrained & biased manager can cause 10/
Men are more likely to receive actionable feedback in performance reviews, whereas women receive vague personality criticisms. Again, guess which is useful for getting promoted? 11/
Research shows people like to hire people like them. You will need to overhaul your interview process: 12/
Also, read these tweets from a former Googler about how the interview process is set up so that one person can easily nix women and people of color. This is true at a lot of tech companies 13/
Another key obstacle to diversity is that HR & Legal are often big perpetrators of discrimination. I'm unsure of how we solve this (other than overhauling the legal system) 15/
Companies don't want the legal liability of admitting there is any defect in their processes, which makes it nearly impossible for them to correct mistakes or improve 16/
To be clear: improving diversity in your org will require substantial changes (to your promotions process, performance reviews, hiring, & culture), not everyone will like those changes, and you will find out that folks you thought share your values actually don't 18/
When people expect that diversity can be improved with relatively "easy" gestures (eg: sponsor Grace Hopper Conference, tutor middle school girls, host a women's panel), they aren't prepared for the conflict that arises from anyone advocating for deeper, meaningful change. 19/
One example: Google had a comprehensive racial justice training that many found effective. Google cut the program so as to "avoid backlash from conservatives" & to avoid "claims by right-wing white employees about Google discriminating against them" 20/
I agree with @karlitaliliana's thread that we need a new set of laws to make workplace racial discrimination actually illegal. Even though I've been writing about what companies should do, this does not replace the huge need for regulation 21/
If you know folks who prefer video, here is a 9 minute video covering some of this material (the need to start at the *opposite* end of the pipeline: the toxic workplace, bias in performance review & mentorship, changing interview process) 23/
• • •
Missing some Tweet in this thread? You can try to
force a refresh
Some folks have asked about data vs. algorithms. Treating these as separate silos doesn't really make sense, and it contributes to a common perception that the data is someone else's problem, an unglamorous & lesser task:
Machine learning often fails to critique the origin, motivation, platform, or potential impact of the data we use, and this is a problem that we in ML need to address.
Q: Is AI development trapped in a paradigm that pursues efficiency above all else? @ResistanceAI
@Abebab cites ongoing work that finds efficiency, accuracy, & performance are the key values mentioned in most ML papers
Q: Is AI development trapped in a paradigm that pursues efficiency above all else?
@red_abebe: Efficient for whom? With example of criminal justice system, is it efficient to have 2 million in USA in prison?
Noopur Raval: The efficiency paradigm can show up in unexpected forms, including many projects claiming to be for social good. Technology can appear part of a mystical, deceptive promise to make things better.
This idea that you can't highlight problems without offering a solution is pervasive, harmful, and false.
Efforts to accurately identify, analyze, & understand risks & harms are valuable. And most difficult problems are not going to be solved in a single paper.
I strongly believe that in order to solve a problem, you have to diagnose it, and that we’re still in the diagnosis phase of this... Trying to make clear what the downsides are, and diagnosing them accurately so that they can be solvable is hard work -- @JuliaAngwin
With industrialization, we had 30 yrs of child labor & terrible working conditions. It took a lot of journalist muckraking & advocacy to diagnose the problem & have some understanding of what it was, and then the activism to get laws changed
I know about diversity-washing, I know about the empty lip-service. But I still can't get past the contrast between @JeffDean's tweets (h/t @EricaJoy) and his treatment of @timnitGebru-- never having a conversation with her, not telling her manager, denying her DEI experience,...
I have long admired @timnitGebru for her brilliance, moral courage, clear voice in speaking up for what is right, & influential scholarship. It is truly terrible that Google would do this.
In this thread, I want to share some of Timnit's work I love
I've quoted "Datasheets for Datasets" (2018) in many of my talks & assign it as reading in my class. It highlights decisions that go into creating & maintaining datasets, and how standardization & regulation came to other industries
Timnit worked with @jovialjoy on the original GenderShades research, which has had a profound impact on facial recognition, led to concrete regulations, and changed our industry
Reciprocity is a key part of life. Surveillance undermines reciprocity. Every time we opt for surveillance or extractive technology, we undermine reciprocity and relationship. -- @doxtdatorb#AgainstSurveillance
Between 1971-1974, a Detroit Police Department surveillance unit called STRESS (Stop The Robberies, Enjoy Safe Streets) fatally shot 24 people, 22 of them African-American @hypervisible#AgainstSurveillance