Charlie Steiner comments on What are the major underlying divisions in AI safety?