work

I was very fortunate to be able to work remotely this week from San Diego! Most of the time, I spent taking my regular meetings, which had to get reshuffled a little bit. Other than that, it was mostly the typical lecture, paper, and grant writing. I am happy though that my coauthors and I were able to almost finish a draft of our STOC paper submission! There's still some editing/cleaning we need to do, but we should be posting it soon (or else, it'll be posted by the conference itself). That leaves me still two weeks to try and get my other project in a more reasonable shape too!

Other than the usual stuff, I was also able to get access to ChatGPT pro through the Stevens SIAI! I've been playing around with it to try and get to to solve some open problems but haven't had much success yet. I want to also see if I can set something up where it interacts with my Claude version together give even better results. For now though, I honestly haven't been too impressed relative to what I've been doing with Fable recently.

non-work

Since I was with my partner, most of my time outside of work was spend with her. We were able to get brunch (many times), go to the beach, get ice cream, and explore more around the marina. There were also a number of nights we just spent in with our cat to play new board games and watch TV shows. It was the first time I'd ever actually played Jaipur and Lacuna both of which were fun!

Toward the end of the week it got pretty hot and felt like summer. In fact, when I flew back on Saturday, the weather was mid 80s I think. But, I came back to 50 degree New Jersey weather, which was a good reminder that it is in fact Fall. I really like the fall weather though, so it didn't bother me.

Overall, the trip was really fun! I'm always happy to spend time with Michelle. I also appreciate that we were able to be together despite her work and marathon training ramping up.

questions

• What is the best way to use GenAI to solve problems?
  • I've seen a recent paper posted which used ChatGPT Dot, but surely there are better ways
  • I realize most of the big results are done by the labs' internal models though, so I'm not sure how valuable using the public models are
  • I also wonder if I should be spending my time on loftier goals than solving these problems. Should I just be doing CS Theory advocacy? Reforming the IACR?
• Which seeds/mysteries remain the cybersecurity community that could in principle be solved?
  • It was cool to see Anthropic verify the honest generation of NIST seeds

Thanks for reading to the end! Here are a few pictures from the week.