generative music algorithms are not something new. brian eno was somewhat of a pioneer in the idea in the 1970s, but his own ideas have roots in the ideas of composers like iannis xenakis in the middle of the last century, which come out of serialism. musicians have been trying to destroy the book on music theory since the day after it was written.

i have used algorithms to create music as well, most prominently on this record, released in late 2000:

https://jjjjjjjjjjjjjjjjjjjjjjjjjjjjjjjjjjjjjjjjjjjjjjj.bandcamp.com/album/deny-everything-lp-4th-record-sample-free-instrumental-remaster

the last track, acidosis, features live performances (piano, guitar, bass, organ, synth, mandolin, flute) and algorithms generating sound via midi sequencing (drums, synthesizers, noise generators) and is very much a collaboration between myself and the computer.

my high school music teacher in the mid 90s was fond of a computer program called "jazz in a box" that would generate midi patterns using ai algorithms. how real it sounded depended on how real the sound fonts were, not on the underlying computer program.

i would be more concerned about midi sequencing + sampling technology taking the jobs of session musicians for instruments outside of standard band instrumentation. as a composer, i can afford to tell a computer to play a violin part, but i can't afford to hire a violin player. maybe i could find a friend that would do it for free. if i was a major production studio, i would be able to afford it but wouldn't want to if i didn't have to. sampling is a more serious threat than ai, but it's also progress in the sense of maximizing compositional flexibility and creative options in real-time spontaneity during the composing process.

https://www.washingtonpost.com/opinions/interactive/2024/music-ai-creativity/?itid=mr_opinions_4