This is exactly why I don't trust LLMs (and therefore why I don't use them). When dealing with something I know about I can see the many mistakes they make - I would have to be a complete fool to trust them to do better on subjects I don't know about.