Evolution without accidents and also no intelligence?

Because of the propensity of AIs like ChatGPT to "hallucinate" and invent corroborating details, I am reluctant to respond, but the arguments seem mostly to be taken from available sources on the Internet including Wiki.  

Indeed, we know wikipedia is unreliable because it contains sceptical bias and omits or misrepresents counter-arguments. In the case of AIs like ChatGPT there is an even more difficult obstacle, 'hallucination', inventions and fiction presented in a style which might appear convincing. That means everything has to be double and triple checked, before even beginning to consider whether it is coherent or representing a legitimate argument.

In mundane topics AI may be useful, in controversial areas I suspect it can only muddy the waters.