Jan Kulveit comments on Risks from Learned Optimization: Introduction