Friendlier AI through politics — AI Alignment Forum