Kind of a selfish question, but I really enjoy doing math puzzles and making an impact, and my general impression is that these two aren’t that compatible. However, it could have a very large impact on my life if this turns out to be false.
FWIW, my current best guesses would be theoretical alignment research (especially ARC’s research and MIRI’s research), forecasting, and maybe some economics work related to global priorities or policy that involves game theory or something like that.
You might be interested in some of the answers to a different question: “What are the coolest topics in AI safety, to a hopelessly pure mathematician?”
This is extremely helpful, thank you.