PeterMcCluskey comments on Can time preferences make AI safe?