Not closely related, but back when I was first reading the book, the idea of computer generated songs sounded like "flying future car" delusions, and learning AI in the early 2000's even confirmed not the impossibility but the crazy limitations of all this.
I have just listened to a podcast last month that mentions how there are songs on Spotify made entirely by AI, and 97% of the people they asked couldn't tell apart regular songs from the AI generated ones.
On one hand, it's remarkable. On the other hand, we're cooked. What's even more depressing is that many-many, even more worrying things are getting pretty accurate in the book. Maybe not back in 1984, but we're witnessing the convergence.
Well, I didn't necessary mean the structure of a song (that doesn't have the complexity that would challenge an AI agent... in the early 2000's, even), but more like coming up with their own lyrics that even make sense, and producing human speech with rhythm and musical tones.