How are OpenAI expected to align a hyper-intelligent entity if they can't even align themselves....