AI-generated music might blur lines between art and algorithm

Written by Danielle Chiriguayo, produced by Bennett Purser

“The cat’s out of the bag. That technology is here … and it's taking off in a much bigger way where it's going to actually affect our lives as people who love music,” says Pitchfork Senior Staff Writer Marc Hogan about AI-generated music. Photo by Shutterstock.

Songs created by artificial intelligence have flooded the internet recently, including “Heart On My Sleeve,” inspired by the music of Toronto-born rapper Drake. Meanwhile, other AI-generated visuals and chatbots like ChatGPT are spreading. Marc Hogan, senior staff writer at Pitchfork, recently wrote about how the tech could dramatically shift the way music is created and consumed. 

Hogan tells KCRW that the technology mimicking voices isn’t new, but “Heart On My Sleeve” shows it’s easy to use and can spread quickly. 

What’s complicated, however, is whether copyrights can apply to these songs. Hogan says in the faux-Drake situation, Universal Music Group was able to take the tune down from streaming services.

“All of that is stuff that the lawyers and the music industry are going to figure out, and hopefully figure it out in a way that is good for artists and good for creators and good for consumers,” Hogan explains. “The cat’s out of the bag. That technology is here … and it's taking off in a much bigger way where it's going to actually affect our lives as people who love music.”

Meanwhile, AI is being used to revive voices of deceased artists, including The Notorious B.I.G.

“There's always that hunger because they're always putting out posthumous tracks by musicians. Some estate or somebody who wants to copyright will have the idea at some point of creating ‘new music’ by a beloved artist, and what will that mean? It's going to be kind of scary.” 

Hogan also points to creators who are making AI versions of classical music. He asks: Once machine learning figures out the algorithm behind music, what does the music become? 

‘It's just numbers. There's no end to it. And sometimes it can presumably be difficult to differentiate which is real and which is fake.”

He adds, “When you come to Drake’s songs, there's a formula to them. His voice is usually pretty processed anyway. In ['Heart On My Sleeve’], his voice does not sound that different. See, I’m talking about it like a person. That voice does not sound so different from the actual Drake voice.” 

At the same time, Hogan says in an ideal world, musicians can harness AI to make the industry more accessible.

“It'd be awesome if these new tools democratize music creation, and let lots of new people who previously wouldn’t be considered musicians do some incredible art that we can't even imagine yet.”

But in good conscience, Hogan says he can’t blindly advocate for this new technology. That’s because it can mean job loss, especially for those in technical roles such as mixing and mastering. 

“This stuff is going to very soon, if not already, allow you to just say, ‘Hey, could you mix this in the style of my favorite mixer or the person who mixed my favorite record?’ And it will do that for you.” 

Some artists have embraced their AI-generated vocal clones, including Grimes. She released a new platform called Elf.Tech, which allows users to create fake Grimes songs by uploading their own audio. 

“Grimes, right after the whole fake Drake saga, said, ‘Hey, anybody who wants to use my voice can, and let's split the royalties from it.’ … It's simple. It's not really Grimes singing. The AI knows what her voice sounds like. It's not 100% how she would sing it, but it’s bizarrely close.”