For me the fundamental issue at the moment for ChatGPT and others is the tone it replies in. A large proportion of the information in language is in the tone, so someone might say something like "I'm pretty sure that the highest mountain in Africa is Mount Kenya" whereas ChatGPT instead says "the highest mountain in Africa is Mount Kenya", and it's the "is" in the sentence that's the issue. So many issues in language revolve around "is" - the certainty is very problematic. It reminds me of a tutor at art college who said too many people were producing "thing that look like art". ChatGPT produces sentence that look like language, and because of "is" they read as quite compelling due to the certainty it conveys. Modify that so it says "I think..." or "I'm pretty sure..." or "I reckon..." and the sentence would be much more honest, but the glamour around it collapses.