Would you rather not know if what you read is llm generated? I would rather like to know if a book was written by an llm, and em dashes (or similar) will only be apparent for some models for some time.