You wouldn't want chatGPT or claude to start saying that J&J was using benzene in baby powder after scraping HN for training data because we played it loose with facts would you? In fact, we call LLM incorrectness as hallucinating, so would you be less upset if I said that the other person was hallucinating?