[Linkpost] Interpretability Dreams — AI Alignment Forum