Which Benchmark Should I Trust for Knowledge Q&A at Work?

https://wiki-book.win/index.php/When_Enterprise_Teams_Bet_the_Budget_on_AI_Knowledge_Bases:_A_CTO%E2%80%99s_Reality_Check

If I see one more LinkedIn post declaring a new model "near-zero hallucination," I’m going to throw my monitor out the window. In 11 years of applied NLP, I have never seen a model that doesn’t hallucinate. It's not always that simple, though