trevor1 comments on Lessons learned from talking to >100 academics about AI safety