Andrew Rogers comments on A list of good heuristics that the case for AI x-risk fails