As a musician I feel like AI is currently destroying music. It won’t generate anything new or truly creative, but damn does it make “good enough” take 10 seconds.
Abso-fucking-lutely. I don’t think it’s quite there yet, it puts too many strange artifacts in the music currently, but it’s getting damn close to “good enough”.
Too many people think the danger is that it’s awful. The danger is that it’s mediocre. Because cheap, easily reproducible, and mediocre beats excellent, expensive, and messy every time because AI doesn’t get tired. It doesn’t go on a bender or get caught doing something reprehensible or burn out. It’s never late. It just sits there waiting to be told what to crank out next.
So you’ve got this thing that can’t move art forward. It can’t inject that one really fucking cool thing in there that changes everything. AI can’t hate a song it’s creating so it emphasizes things in a weird way.
Compare it with Max Martin productions. He didn’t invent manufactured music but he created a hell of a pipeline for folks to rhyme fire with desire. But even that relies on people who can sing with the timidity of youth and the confidence of a person who has been told the world is theirs. Or someone with no real understanding of a song singing it in a way that gives it a different meaning than the original intent. Or someone barely hanging on and pouring their entire person into their performance because they have nothing else.
AI can’t do any of that. It can’t turn a word into a god damned grenade. It’s going to remix everything that came before. Not in new and exciting ways. Not in thought provoking ways. But in algorithmic ways. It’s flat. The lyrics will tell a story that resolves. The rhymes will be perfect. You won’t get a banjo in metal or a calliope in video game music unless it’s a game about a clown. It’s not going to give you soul and wit. It will give you a snapshot of where music has been and is up to the point of its last training data.
I have an entire tangent about how it’s being used politically currently (go look up Danny Bones) and how it does not get tired or embroiled in controversy and being “good enough” makes it the perfect propaganda machine. But that’s for another day.
I’ve been really enjoying just playing my acoustic instruments, not bothering to record anything at all. I think campfire sings are going to make a major comeback.
It depends what your standard is for good enough, honestly. If you’re making art for the art, I certainly haven’t seen anything from AI music that seems actually artful, and I wouldn’t feel comfortable putting my music out into the world that didn’t feel like art to me. If you’re churning out 30 second jingles for advertising or microtransaction-riddled mobile games, maybe then the standard is lower
E: of course, if “good enough” is “enough to make a C suite douchebag give it the okay with dollar signs in their eyes” then yeah.
As a musician I feel like AI is currently destroying music. It won’t generate anything new or truly creative, but damn does it make “good enough” take 10 seconds.
Abso-fucking-lutely. I don’t think it’s quite there yet, it puts too many strange artifacts in the music currently, but it’s getting damn close to “good enough”.
Too many people think the danger is that it’s awful. The danger is that it’s mediocre. Because cheap, easily reproducible, and mediocre beats excellent, expensive, and messy every time because AI doesn’t get tired. It doesn’t go on a bender or get caught doing something reprehensible or burn out. It’s never late. It just sits there waiting to be told what to crank out next.
So you’ve got this thing that can’t move art forward. It can’t inject that one really fucking cool thing in there that changes everything. AI can’t hate a song it’s creating so it emphasizes things in a weird way.
Compare it with Max Martin productions. He didn’t invent manufactured music but he created a hell of a pipeline for folks to rhyme fire with desire. But even that relies on people who can sing with the timidity of youth and the confidence of a person who has been told the world is theirs. Or someone with no real understanding of a song singing it in a way that gives it a different meaning than the original intent. Or someone barely hanging on and pouring their entire person into their performance because they have nothing else.
AI can’t do any of that. It can’t turn a word into a god damned grenade. It’s going to remix everything that came before. Not in new and exciting ways. Not in thought provoking ways. But in algorithmic ways. It’s flat. The lyrics will tell a story that resolves. The rhymes will be perfect. You won’t get a banjo in metal or a calliope in video game music unless it’s a game about a clown. It’s not going to give you soul and wit. It will give you a snapshot of where music has been and is up to the point of its last training data.
I have an entire tangent about how it’s being used politically currently (go look up Danny Bones) and how it does not get tired or embroiled in controversy and being “good enough” makes it the perfect propaganda machine. But that’s for another day.
Modern elevator music.
You said more than I did in my whole fucking rant. That’s exactly right.
I’ve been really enjoying just playing my acoustic instruments, not bothering to record anything at all. I think campfire sings are going to make a major comeback.
It depends what your standard is for good enough, honestly. If you’re making art for the art, I certainly haven’t seen anything from AI music that seems actually artful, and I wouldn’t feel comfortable putting my music out into the world that didn’t feel like art to me. If you’re churning out 30 second jingles for advertising or microtransaction-riddled mobile games, maybe then the standard is lower
E: of course, if “good enough” is “enough to make a C suite douchebag give it the okay with dollar signs in their eyes” then yeah.