Roman Leventov comments on Towards a solution to the alignment problem via objective detection and evaluation