Utility, probability and false beliefs — AI Alignment Forum