Exactly. Most wrong-on-the-internet people have tells. And most of the rest have history. We all use this all the time on HN. But with LLMs the former gets engineered away and the latter is nearly useless. Except of course to the extent that we say,"Wow, ChatGPT is absolutely untrustworthy, so we should never listen to it." But given how many people are excited to use ChatGPT as a ghostwriter, even that is limited.