This of course is the big question, thank you for asking it. Whether humans actually have free will is a rich discussion for the philosophers. I beleive that humans have free will bounded by the laws of nature as described by Quantum Field theory. So while I can choose when to load my dishwasher I can’t telekinetically send the forks flying over there. As best as I understand, I decided when and how to load the dishwasher. So, as far as I am concerned within my realm, I have free will.
I think it will be similar in AI. Systems will make choices based on the realm of inputs designed for it. How it makes the decisions will no longer be based on human-coded rules, but on its own ever-growing self-coded rules. We may not be able to discover how they arrived at their decisions. In essence it could appear to be free-will. The question is there some scenario where a system could make decisions that are outside of its realm, for example, seek other inputs via the Internet to enhance its decision-making ability? I’d love to hear experts opine.