Technical AGI safety research outside AI — AI Alignment Forum