Promoted to curated: Overall this seems like an important and well-written paper, that also stands out for its relative accessibility for an ML-heavy AI Alignment paper. I don’t think it’s perfect, and I do encourage people to read the discussion on the post for various important related arguments, but it overall seems like a quite good paper that starts to bridge the gap between prosaic work and concerns that have historically been hard to empirically study, like deceptive alignment.
Promoted to curated: Overall this seems like an important and well-written paper, that also stands out for its relative accessibility for an ML-heavy AI Alignment paper. I don’t think it’s perfect, and I do encourage people to read the discussion on the post for various important related arguments, but it overall seems like a quite good paper that starts to bridge the gap between prosaic work and concerns that have historically been hard to empirically study, like deceptive alignment.