Prospects for Alignment Automation: Interpretability Case Study — AI Alignment Forum