> (The gp's comment is an example of how chatbots hallucinate because they train on the text of people unintentionally hallucinating.)
We're now applying LLM anthromorphism back on people...sigh
We're now applying LLM anthromorphism back on people...sigh