Super proud of this work
https://t.co/QpDNcy5mQU
much of the world needs their own input on what model behavior is, esp groups whose values are being overwritten.
What a joy to watch a panel full of incredible women, @navrinasingh@mmitchell_ai@CummingsRenee, talking about how to make the Responsible AI movement more inclusive. Join the @CredoAI Responsible AI Summit now to watch, you don't want to miss this: https://t.co/Uj20Vmme2P
@mmitchell_ai Such an important point! When I first started talking to banks about discriminatory bias in credit models in 2017, many believed that excluding protected attributes from their inputs meant their models weren't biased. We've come a long way but still have a LONG way to go.
Cathy O'Neil mentions a 2x2 matrix to help people understand ethical issues with their work: Stakeholders x their concerns.
So I'd like to offer my complementary matrix here, to help think through what the concerns are.
"When you talk about AI & data, you're dealing with some of the most fundamental civil rights in America -the ability to enter and thrive in the workforce and not be discriminated against based on protected characteristics." @KSonderlingEEOC @USEEOC@shshattuck#CredoAISummit2022
I had such a wonderful time chatting with these brilliant folks about how we can better build bridges between technical, business, and policy stakeholders—excited to share with all of you later this week at the @CredoAI Responsible AI Summit!
Ensuring that AI systems are responsibly developed, deployed, and used takes a village. Hear Cathy O'Neil @KSonderlingEEOC, Professor David Danks @UCSanDiego and @shshattuck@CredoAI discuss this and more at the 2022 Global Responsible AI Summit https://t.co/rmiwQHin4U
@CitiBikeNYC@lyft This is particularly dumb because I have been using @lyft for the last 8.8 years. Anyone looking at my profile in your database would see that my age cannot possibly be right, how could I have requested rides before your system thinks I was born?
I mis-typed my birthdate when signing up for @CitiBikeNYC, and now across Citi Bike and @lyft, I can't use my account because you think I'm under 18. A very simple thing to fix in a DB somewhere. Impossible for me to fix through your stupid FAQ support system.
I didn't see the point of image generation models like #Imagen and #dalle, but now I do: they can help people *see* model biases that are hard to explain with words (and even formulas!)
We need a book like "It's So Amazing!" but instead of teaching kids about where babies come from, it teaches your uninformed family members how to engage in debates about AI sentience.