>It seems to me something like history might be more at risk. In mathematics a wrong result can be checked. With something like history checking answers is not always so easy.

Isn't that exactly why we need historians? How do you fact check an interpretation of evidence or of history? I would never trust a chatbot for its take on history, and that's partly due to the fact that we've seen how political it can get. In the worst case, the communication of history can be used to misrepresent cultures and limit freedoms. How we interpret history informs our present day policies. There's no way we'll ever leave that to a computer.