[mplewis]

LLMs don’t answer based on truth. They are a glorified autocomplete trained on a large corpus of text.