Joe Collman comments on Asymptotically Unambitious AGI