Gordon Seidoh Worley comments on Considering the Relevance of Computational Uncertainty for AI Safety