Neerja en España #11: FAT*
Originally written February 2020
I closed out the insane month of January with a 4-day trip to Barcelona for the FAT* conference, which pulled together computer scientists, lawyers, people in the tech industry, and social scientists to talk about fairness, accountability and transparency in computing. This was my first academic conference ever, and I had an incredible time!
I decided to go to the conference because it was conveniently a 1-hour train ride away, Fulbright was going to fund my attendance, and it seemed important and relevant to my interests. It ended up being much more meaningful than I had anticipated - I’m so happy I had the opportunity to go! Beyond just talking to CS academics and people in the tech industry, I got to talk to lawyers and professors in non-CS fields. There were so many cool talks that exposed me to ways of thinking I had never experienced before. There were also a handful of Berkeley PhD students there and it was great to meet them and learn about ways I can be involved in similar discussions once I go there. I’m definitely excited for what’s ahead!
I’ve always understood that thinking about the implications and ethics of CS is important, and put effort into educating myself in those areas, but I’ve never learned as much about bias in computing, or how to make AI explainable, or challenges in creating AI policies, as I did during this conference. The talks I attended definitely changed my personal evaluation of different AI systems I encounter - right now, neural networks optimize for some loss function that minimizes error, but what they instead were optimized in a way that penalizes bias? I also thought AI explainability was an extremely interesting and important area, since frequently a neural network will give us a result and but we have no idea why it is returning that result, and as a result neural networks lead to all sorts of problems (read this paper about how a single pixel change can completley fool a neural network).
Beyond the conference, being in Barcelona and with the other Fulbright researchers was incredible. There’s a cohort of about 10 Fulbright researchers in Barcelona and they are all amazing people. We had a lot of fun group dinners in the 3 nights I was there, including one with most of us crammed into the corner of a tapas bar and being unnecessarily crazy after one round of beers. At some point in this visit, the Barcelona crew officially adopted me as one of their own, adding me to the group chat and everything, which was adorable. Audrey, a neuroscience researcher who is going to go to med school next year, very graciously hosted me the entire time I was there, and we bonded incredibly. It’s rare that you can spend 6 hours one on one with someone, have nonstop conversation the whole time, and then be shocked that 6 whole hours have passed. I attended the conference with Anna and Miranda, the two other CS researchers (both of who are so cool and brilliant and wonderful), and while none of us have explicitly done research in AI fairness or related fields, we had a great time exploring the conference together and talking about how our views on our own research changed throughout it.