Nice post. I studied Electronic Music Technology for three years in the early 80's largely as the result of Kraftwerk, Tangerine Dream, and dare I say it, Georgio Moroder (as well as the BBC Radiophonic Workshop)! For me, while at the time the technological achievement was huge, it was clear that by the early 90's pretty much any idiot with a synth could do what those guys had done. Does that lessen anything?
I think it's like this....
There are probably 4-year olds who can play it now, but man, so what. In it's moment it was glorious, every 14-year old kid on the planet wanted to be able to play it. I listen to Kraftwerk today and love the fact that you can hear the hiss as they turned on a machine, various clicks and thumps and other random noises, the odd cough or grumble. My acoustics lecturer told me once that the test of a good orchestral recording was whether you could hear the pages being turned, i have always felt that. As is so often the case, it's the errors/weaknesses/nuances that make the art - the 5/4 bar in Rain, the mic stand falling over in Long Distance Love, Bonzo's squeaky pedal - OK they're extreme examples but the reality is that when music is made by humans in real time you can tell. AI can probably do that, but it probably won't because the people who'd use AI to make music would not see music that way.