Why language models hallucinate | Not Hacker News!