x
Taking LLMs Seriously (As Language Models) — LessWrong