moving to san diego
I've decided to stop including a "things i saw" section. This is mostly because I already have a public JSON feed with many things I go through online. It's tedious when I want to copy something over here. I'll miss things that I see and aren't in my RSS feed, but cé la vi.
This post has been delayed while I was in San Diego this week.
work
I was basically on a pseudo-vacation while in San Diego. But of course, with the looming Eurocrypt deadline, I couldn't take a full vacation. Having to balance between vacation activities, moving and writing honestly helped me focus a lot. I was able to just have a few concentrated hours of writing and productivity each day that were ruthlessly prioritized toward the coming deadlines. Next week, I'm excited that I'll have a lot of progress to report and continue finishing up before the deadline in less than two weeks (!).
Unfortunately, this prioritization probably cost me in other areas of job. I was hoping to get the rest of my office equipment set up, but almost put that on hold for the trip. As far as I can tell, I didn't miss anything major and will still be around for faculty meetings and stuff next week, so I pretty much just had a really packed and productive week.
non-work
This was a very exciting and stressful week for me. My partner is about to start a new job in San Diego for a while (congrats to her!). But that involved moving her across the country this week. That's pretty much what occupied the latter half of the week. A lot of furniture building and learning the area she's living in. She's in a nice community in a nice party of San Diego, though, so I'm looking forward to all my future visits with her.
We tried a lot of really great food, got to visit La Jolla, and ran along the harbor in the area. It was a really good balance of vacation and moving, which makes me glad that we spread it out across half a week. In the future, I'm hoping to be able to visit for longer and maybe call it a work trip by talking to people at UCSD. For now though, I'm happy with the week that we had.
questions
What would companies be wiling to sacrifice for more security around their AI applications?
This is intentionally pretty open-ended. I've been thinking about the trade-off of different private inference/training mechanisms, but I don't know what (if any) realistic trade-offs companies would be willing to make.
How important is it for cryptographers to be involved in AI security and policy?
I feel like a lot of AI policy discussion is beyond my area of expertise. But, at the same time, I think some percentage of AI policy debate could benefit from concrete threat models and adversarial modeling
Thanks for reading to the end! Here's a picture of the La Jolla sunset:
