[Linkpost] Scott Alexander reacts to OpenAI's latest post — LessWrong