Notes on Self-Control — AI Alignment Forum