to think more clearly about AI, people should anthropomorphize more or less ?