Richard_Kennaway comments on Problems in AI Alignment: A Scale Model