Research update: ALTER is funding my research!
Ram Rachum
Hi everyone!
I'm delighted to announce that ALTER has agreed to fund my research for the next 6 months!
I've been looking for a sponsor for a year now, as I was burning through my savings at an alarming rate, so I'm happy to have found ALTER. They are a new foundation in the Effective Altruism space, funding Israeli researchers in the areas of AI Safety, pandemic preparedness, health security and progress studies. Follow ALTER on Twitter.
Now I'll be able to focus on my research, putting aside financial concerns until next year.
Retrospective on goals for last month
In last month's update I outlined a list of goals. Here's my update on these goals:
Prepare the Stubborn talk for the RaD-AI workshop: ✅ Done
My talk is all prepared. I practiced it on the masters students in the GOLD lab, and it was well received. I did get feedback that I'm spending slightly too much time on motivation, but I think it'll be okay.
Figure out the strategy for my first full-length paper: ✅ Done
After consulting with my researcher friends, I decided that I'll start writing the paper without aiming for a specific conference. I'll write it as if I'm aiming to just put it up on arXiv, and after it takes form, I might decide to adapt it to the requirements of conferences such as AAAI or AAMAS.
More about that paper below.
Stuff I've done beyond the monthly goals
Writing the dominance hierarchy paper
I'm having a lot of fun writing my paper. Previously I used the term "social order" to talk about the order in which agents play hawk to each other, but after reading up on Wikipedia and consulting researchers, I concluded that a more acceptable term would be dominance hierarchy, so that's what I'm using.
I first wrote the abstract and around 2 pages that would go in the middle of the paper. I showed the paper to Reuth and she tore it a new asshole. We had to split our feedback session over two meetings, and the notes I took were almost as long as the paper. I'm still struggling to understand how academic papers are written. I'm going in the right direction, but it'll take a while before I'll really get it. Meme.
After these feedback sessions I've written a couple of pages of introduction and definitions, and now I have a total of 5 pages including citations. I'm guessing the paper will be around 10 pages long, though I may pivot to a different paper style so who knows.
ChatGPT has been very helpful in my writing process. Sometimes I write a paragraph but there's a bit that doesn't sound good, so I can copy-paste it into ChatGPT and ask it to rephrase. It offers me several different options. I sometimes tell it I want to express something and it writes paragraphs with great ideas, including references! I even have it generate LaTeX math for me. It reviews the prose that I write and it gives me feedback... It's amazing. I've been using ChatGPT for lots of stuff, but its competence in paper-writing is especially high.
While I'll continue working on the paper, it's not an official goal for the upcoming month, because I'll spend half that month on vacation.
I've also been spending some time diving into the existing literature for dominance hierarchies in animal life. A few papers I've read so far are:
I was accepted to the CAIF Summer School!
Last month I got an email about CAIF organizing a summer school. I've been following this foundation for a year now, so I immediately signed up. A few days ago I got the acceptance email, and I'm very happy about that. (It's funny that the way they worded their email made me immediately think that I was rejected. They wrote "Dear Ram, Many thanks for your application to the Cooperative AI Summer School" and I'm thinking they're buttering me up before they apologetically say I was rejected... But I was pleasantly surprised.)
There are gonna be talks by Joel Leibo, Vincent Conitzer, Jakob Foerster and other top-notch researchers. I'm also looking forward to meeting a few friends from the Berkeley MARL reading group. Unfortunately, not everyone I wanted to meet was accepted.
I hope I'll get a chance to present my dominance hierarchy experiments. I know I'll get a chance to do a poster session, but I hope I could also get a slot for a talk.
CAIF will be covering the cost of my participation, which I'm grateful for.
My goals for this month
Co-chair and present at RaD-AI workshop.
After over six months of preparations, it's finally happening. On the 30th I'll be co-chairing the RaD-AI workshop, and presenting our paper there.
This'll be my first time attending a workshop in-person, and also my first time co-chairing a workshop, and also my first time presenting at a workshop in person... I think it's gonna be fun, though I still feel I should be intently listening to the social cues to figure out how people behave in these events.
Attend AAMAS.
Right after the workshop, I'll be attending AAMAS core days, which will start on Wednesday and end on Friday. I'm excited, though I think that my brain will melt after 3-5 talks. I hope to enjoy the mingling part. If you're at AAMAS, hit me up.
When I signed up for AAMAS, I saw there was an option to sign up for the banquet. So apparently that's a thing. I'm not completely sure how formal it's supposed to be. But I like dressing up and consuming alcohol, and I don't get many opportunities to do that in Israel.
Have fun in Lisbon.
After a few more days in London, I'll be flying over to Lisbon for a vacation! I've never been to Portugal so I'm excited. I'll be there for 2 weeks, hit me up if you're around. I decided I'll try staying in a hostel for at least a few days, so I could feel like a young adventurous person for a bit. It's possible I'll move to a hotel if the magic will wear off too quickly, or if I'll have roommates who snore louder than me.
That's it for now. See you next month!
Ram.