wrapper-minds are the enemy — AI Alignment Forum