I just wrote a piece for the New York Times on A.I. music: the companies making it and its potential implications. You can read it here.
It’s a strange area to look into as, every moment, you’re stuck between thinking, ‘It’s so cool people are working on this’, and, ‘What on earth happens if they succeed?’ The questions it raises for music’s future are almost overwhelming.
The dilemma was summed up by these quotes that originally ended the piece (they had to be cut due to space):
“I think people will accept [A.I. music],” said Margaret Schedel, co-director of computer music at Stony Brook University, who has been observing the field for over twenty years. I mean that in all contexts – on the radio, in shops, everything. There’ll be some initial resistance, then it’ll become ubiquitous.”
“The reason I like computer music is hopefully it can go beyond what we as humans can,” she added. “That’s the exciting thing. The sad thing is the potential automation and putting musicians out of work.
“But don’t put that in your article as then the A.I. people will come and get me.”
There is one style of music, though, that I think is ripe for A.I.: national anthems.
Given there are only a couple of hundred of them, and that most share similar a similar musical style and lyrics, surely someone at Google Brain’s Magenta project or DeepMind could quickly knock out a programme to learn from that source material and write one? It might be an improvement.
If you’re a new country looking to get some cheap publicity, it may be worth you contacting some of the companies mentioned in the article!