This just seems like a longer version of the HN comments I see in every single AI related post. No such thing as a hallucination, LLMs don't really "understand"

This just seems like a longer version of the HN comments I see in every single AI related post. No such thing as a hallucination, LLMs don't really "understand" therefore they can neither say truths nor lies just like autocorrect can't be honest or dishonest, it "just" chooses a token from a probability distribution, etc. I see this every day. I'm pretty sure most of us who have even a slight interest in AI know the gist of how LLMs work. I'm not sure about what difference it makes in practice.