Gauging interesting for MIRIx

[Vadim Kosoy]

Gauging interest. We had a few MIRIx sessions in Tel Aviv previously, and I'm thinking of reviving that endeavor, but in a different style. Specifically, I want to (i) focus mostly on the learning-theoretic agenda and its vicinity (ii) try to make actual progress (as in, prove non-trivial theorems) rather than only discuss papers. This is a rather ambitious goal since it requires that people come who have the necessary mathematical background and skills, and moreover they should be willing to work on those problems outside of the meetup itself (otherwise chances of making progress are slim). On the other hand, if there are 2-3 such people in addition to myself this is already a critical mass IMO.

If you are interested in this in any capacity (even just to listen in on the discussions), please participate in the Facebook poll ("contribute" means "contribute to making tangible mathematical progress on the relevant problems"). If you don't have Facebook, feel free to reply here. Don't be too conservative in evaluating your own skills: if there is enough interest, I will make another poll later to assess the mathematical knowledge of participants in more detail.

If you know people outside this group who might be relevant to this project, please bring it to their attention.

* "MIRIx" refers to meetups around the world whose goal is making independent research into AI alignment, loosely associated with MIRI.