Machine learning is used for generating music. And it is used to evaluate it too [1]. And there are some insanely cool learning capabilities in it too. Sounds obvious to lock these three algorithms in one room for a week and see what comes out.
Realistically, Learning will find a weak point in Evaluation then drive Generation to optimise this loophole as far as possible and what we’ll get is an unbelievably horrible track.
But in my happy fantasy (or after enough iterations of humans patching Evaluation) it produces something worthwhile. Maybe even better, an absolute hit comes out which everyone is then spending whole days listening. Or not everyone but only a group of people who’s tastes are similar to those of Evaluation (its developers?).
Previously I thought that after all human jobs are eliminated thanks to automation we’ll be left with plenty of time to do art and science. Like rich aristocrats of ancient and medieval times when they were inclined to. Now I think that computers will be better at literally everything. And I won’t listen to music composed by humans anymore. I will listen to works of algorithms, works so beautiful like I’ve never imagined I would love music so much. I keep trying to imagine myself enjoying a live performance by humans, but their own works will sound dull, and them playing algorithmically composed music will feel wrong.
The only thing that I still cannot see computers doing better than me is experiencing my own life (or I fear to imagine that). So the only life left for me will be sleeping and enjoying everything computers make for me. From time to time I could make something for myself, just to feel myself utterly incapable of anything. There’ll be some hipster trend to live on human-made products and services; very spartan-minded people will participate I suppose.
Will anyone be giving birth to children to live in this utopia? I don’t know.
[1] I’m stretching things a bit here. At the moment evaluation is based on opinions of real people. What we still lack is purely isolated evaluation based on the composition itself.