AI alignment risk is real but where i depart from the rationalists is that I have no faith slowing down AI researchers will allow us to solve it via pure deduction in the interim. especially seeing as nobody seems to care
Conversation
Replying to
AI alignment requires iteration. Fail fast, I say