A very thought provoking question, Bob, and one which is difficult to answer with any certainty. The reasons we create music are deeply interwoven with the personal psyche and the desire to implant highly individual perceptions on abstract sound. There is also the physical component, the pleasurable experience of generating music after the manner of a yoga. Whatever AI does, I feel most people are likely to want to preserve these two individual, participatory aspects. While the products of code such as that of David Cope are certainly admirable in a loose, imitative sense, the creative human mind is likely to want more than emulation of music of the past. AI cannot, obviously, substitute for the physical experience, the joy of playing an instrument. It might perhaps evolve sufficiently to partner the brain in the creative act, in much the same way as computers test hypotheses for mathematicians and facilitate otherwise unreachable proofs, the four colour theorem being a case in point. But art is not quite the same as science, it has no universal validation. The solution of a mathematical problem is invariant over all human brains, a work of art has as many meanings as there are perceiving minds.
Technology in the broad sense has, of course, already changed how we record and listen to music in ways too numerous to list, and has done so very quickly. It is not inconceivable that AI might extend that process into the realm of human intellect and emotion. Maybe, instead of our using it, it will commence influencing us. J. B Priestley remarked that all we would have to do to effect this switch is to simply start preferring computer music to our own.