Rafael Harth comments on Is the argument that AI is an xrisk valid?