A.I. is tool that can primarily improve speed and accuracy in completing a task. In one medical field study, ChatGPT outperformed doctors in diagnosing medical conditions from case reports (i.e., improved accuracy), and did it quicker (i.e., speed). In my technical/engineering field, we are considering an internal A.I. tool based on our 60 years of documented experience.
Therefore, yes, I fully expect A.I. could be used to generally improve audio designs but there will still need to be a designer with a personal vision who can interpret the A.I. results and utilize them to avoid pitfalls and possibly learn from what others have done previously. Somebody has to listen to prototypes and decide what the final product will sound like. Also, there is probably enough misinformation related to audio that even A.I. will not get it right all the time.