I don't think we'll get there with AI unless we get to point of creating general intelligence. The AI compositional developments are pretty wild, but I think w

I don't think we'll get there with AI unless we get to point of creating general intelligence. The AI compositional developments are pretty wild, but I think without the varied high-entropy input that comes from lived experience I'm not sure they'll ever really get there. I've repeated this before—many (maybe most) great artist will tell you that the songs didn't really come from themselves—they just sort of felt it, they were handed it, or they reached out and pulled it down. Some of my favourite songwriters are quite open about their process including sitting on the couch noodling mindlessly and playing with melody like it was putty (rather than an equation). Ultimately I think it follows the old adage that if you analyze something too much it will lose another aspect of itself that isn't analyzable. You look to solve the problem of creating great music by reducing it to a problem that we have mathematical models for, then you will lose all of the input from areas we where don't yet have mathematical models. We "don't yet know what we don't know" about music. Definitely a fun area of discussion. I get bored fast when people try and reduce composing to such restricted values, though—except as a compositional technique unto itself. Music is math, alright, in as much as everything is math. But our mathematical model isn't advanced enough to coherently describe the experience in so few terms as people in attempted discussions about objective merits of/in music tend to reduce it.