The point would be to have an interesting and novel experience in an experimental medium - which has been a major driver of art since its beginning.
Also, realistically, most people want entertainment, not art (by your definition). They want to consume experiences that are very minor variations of on experiences they've already had, using familiar and unsurprising tropes/characters/imagery/twists/etc.
The idea that only humans can make that kind of work has already been disproven. I know a number of authors who are doing very well mass-producing various kinds of trashy genre fiction. Their readers not only don't care, they love the books.
I suspect future generations of AI will be better at creating compelling original art because the AI will have a more complete model of our emotional triggers - including novelty and surprise triggers - than we do ourselves.
So the work will be experienced as more emotional, soulful, insightful, deep, and so on than even the best human creators.
This may or may not be a good thing, but it seems as inevitable as machine superiority in chess and basic arithmetic.
I agree with the sentiment that "most people want entertainment, not art," or at least they do a lot of the time. I have a pretty wide definition of what is art, in that almost anything created by a human could be appreciated as art (whether that's a novel, a building, the swinging of a baseball bat, or even a boring sidewalk). But a lot of people, a lot of the time engage with movies and books and the like as merely "entertainment." There's art there, but art is a two-way interaction between the creator(s) and the audience. Even in the pulpiest, most corporate creations. I'm not engaging with cat food commercials as art, but one genuinely could. I agree that AI can generate stuff that is entertaining.
"The idea that only humans can make that kind of work has already been disproven." That I disagree with, and it ultimately is a matter of "what is art." I won't pretend to offer a full, complete definition of what is art, but at least one aspect of defining what is and is not art is, in my opinion, whether is was created by a human or not. There is at least some legal precedent that in order for a copyright to be granted, the work has to be created by a human being: https://en.wikipedia.org/wiki/Monkey_selfie_copyright_disput...
"I suspect future generations of AI will be better at creating compelling original art because the AI will have a more complete model of our emotional triggers - including novelty and surprise triggers - than we do ourselves."
Again, by my definition at least, AI cannot create "original art." But I'll concede that it is conceivable that AI will generate entertainment that is more popular and arousing than the entertainment of today. That is a rather bleak future to imagine, though, isn't it? It seems reminiscent of the "versificator" of 1984.
Also, realistically, most people want entertainment, not art (by your definition). They want to consume experiences that are very minor variations of on experiences they've already had, using familiar and unsurprising tropes/characters/imagery/twists/etc.
The idea that only humans can make that kind of work has already been disproven. I know a number of authors who are doing very well mass-producing various kinds of trashy genre fiction. Their readers not only don't care, they love the books.
I suspect future generations of AI will be better at creating compelling original art because the AI will have a more complete model of our emotional triggers - including novelty and surprise triggers - than we do ourselves.
So the work will be experienced as more emotional, soulful, insightful, deep, and so on than even the best human creators.
This may or may not be a good thing, but it seems as inevitable as machine superiority in chess and basic arithmetic.