romeostevensit comments on RFC: Philosophical Conservatism in AI Alignment Research

[romeostevensit]

asymmetry in the penalties for type 1 vs type 2 errors.