Maxime Riché comments on Which types of AI alignment research are most likely to be good for all sentient beings?