Robustness to fundamental uncertainty in AGI alignment — AI Alignment Forum