whpearson comments on A model I use when making plans to reduce AI x-risk