Rob Bensinger comments on Some abstract, non-technical reasons to be non-maximally-pessimistic about AI alignment