Resolving human values, completely and adequately — AI Alignment Forum