x
Consequentialist preferences are reflectively stable by default - History — AI Alignment Forum