On Doing the Impossible — AI Alignment Forum