How AI Is Changing Music

On Nov. 2, a brand-new Beatles song called “Now and Then” hit streaming services. It features contributions from all four of the band’s members, in spite of the fact that John Lennon and George Harrison died decades ago.

Almost as highly publicized as the song’s existence itself is the fact that it was made possible thanks to AI, which was able to split John Lennon’s original 1977 demo of the song into individual tracks that could then be mixed and mastered. That work, oddly enough, is one of the more straightforward contributions that AI has made to music so far.

Look around the internet for long enough, and you might stumble upon Lana Del Rey singing Phoebe Bridgers’s “I Know the End,” Kanye West covering Taylor Swift’s “You Belong With Me,” or Drake rapping to Ice Spice’s “Munch.” You might also find a collaboration between Drake and The Weeknd, or the Notorious B.I.G. performing Tupac Shakur’s “Hit ‘Em Up.” All these songs, of course, were never actually recorded by the aforementioned artists. Yet you can listen to each one of them online along with hosts of other collaborations, covers, and tracks that were never actually recorded by a living being, thanks to the strange and rather terrifyingly powerful union of music and AI.

Perhaps even more unnervingly, AI-generated music is now well on its way to breaking into the mainstream. In a Sept. 5 New York Times interview, a rep for the TikTok creator Ghostwriter revealed that “Heart on My Sleeve” — a song that uses the AI-generated voices of Drake and The Weeknd — had been submitted to the 2024 Grammys for best rap song and song of the year. Due to the Recording Academy’s guidelines, which specify that songs written in partnership with AI are eligible for Grammy consideration, it seemed like the song might actually make it into the competition.

Grammys CEO Harvey Mason Jr., who initially told The New York Times that the song was “absolutely eligible,” backtracked days later. “Let me be extra, extra clear: Even though it was written by a human creator, the vocals were not legally obtained, the vocals were not cleared by the label or the artists, and the song is not commercially available, and because of that, it’s not eligible,” he said in an Instagram video.

Still, the fact that a song that uses AI-generated vocals was nearly fair game at the Grammys shows just how far AI-made music has come, and hints at how far it might still go. Today, TikTok is rife with viral AI-generated tracks, which range from generally affecting (if morally questionable) to completely absurd. Plus, multiple publicly available apps — such as Endel and Google’s aptly named AI Music Generator Song Maker — now allow users to create mashups of songs with a few clicks. One thing is clear: like it or not, AI and music is a union that’s here to stay.

AI-influenced music has become so prominent that giants like Universal Music Group and Spotify are taking notice. As of August 2023, per The Guardian, Google and Universal were negotiating a deal regarding how to license artists’ voices for use in AI songs; the deal will most likely allow copyright owners to be paid when their voices are used.

AI is, of course, capable of composing music, writing lyrics, generating entirely new vocals, and much more. Naturally, that can be terrifying to hear, especially in a world where most musicians already struggle to make a living with their art.

However, many artists and thinkers don’t necessarily see AI as the foremost threat to musicians at large. Grimes, for example, has openly embraced AI, inviting artists and fans to use her vocals to create new songs, and allowing creators to equally share in the profits from any tracks she approves.

Claire L. Evans, the lead singer of the band Yacht, has also been making AI work for her for years. In 2016, she and her band began working with AI to craft an album, using machine learning to create song lyrics and melodies based on their older music. The product, an album called “Chain Tripping,” dropped in 2018.

Evans prefers to see AI as a tool like any other instrument or plug-in, not a replacement for human creativity. “I think something we realized really early on was that you can’t just take the output as is and call that art. You have to take that as part of the process and figure out how to deconstruct it, how to react to it, how to assemble it, kind of like putting a puzzle together into something meaningful and interesting,” she tells POPSUGAR.

Jason Palamara, PhD, an assistant professor of music technology at Indiana University-Purdue University Indianapolis, feels similarly. He also believes that while AI can create music at a high level, it’s not yet able to emulate the aspect of choice and surprise that characterizes so much of human creativity. AI can emulate a Nirvana song, for example, but it can’t yet innovate in the way that a living musician would. “If Kurt Cobain and Nirvana had continued on to modern day, for all we know, Cobain would be making bluegrass music,” he says.

Still, theoretically, he admits, AI could acquire that ability; after all, it’s growing exponentially almost on the daily. In the years since Yacht released “Chain Tripping,” Evans has also been amazed at the speed with which AI has developed. “We’re having an invention-of-photography-level event in AI development every few weeks. Every month, it seems like these paradigm-shifting technologies are arriving,” she says. “They’re arriving faster than we have the capacity to metabolize them.”

Dr. Palamara also acknowledges that there will be lots of growing pains as AI becomes more prominent in the music world. “I think in the short term, you’re going to see a lot of cringey things like cultural appropriation happening, and it’s not going to be policed in any kind of way,” he says. Both he and Evans say they want to see changes made to copyright laws, which Dr. Palamara notes are already far out of date anyway. Artists should always be able to own their own vocals, he says, and should generally be paid a lot more for their work. He also sees complexities potentially arising when it comes to who owns an artist’s voice or persona after their death.

Still, he notes that while AI could potentially threaten some musicians’ livelihoods, it’s not like high-paying jobs for musicians are plentiful at the moment. “It’s very difficult to make money as a live act, as a songwriter, as a beat maker, as an audio engineer or producer or studio. Someone in the world is making money on music, and it’s not people at these levels, and that’s a problem,” he explains. “I don’t really see how AI music is going to necessarily make this so much worse.”

For now, he says, he would love to see musicians and artists more involved in creating AI. “I do think that if we were, as a musical community, to engage more with AI, we could perhaps steer things in the direction of improving things for ourselves, because we’re already in a pretty tough situation,” he says. Instilling ethics in AI is arguably one of the most important tasks of our time, and we may only have a limited window of opportunity to do so, so the fact that AI is being created by people who often have no connection to the people whose lives will be changed by their products is a huge issue.

That’s why it’s so important to instill ethics into our flesh-and-blood leaders and systems as well. Evans is hesitant to fall into fearmongering about AI when the real threat to musicians and artists often comes from an all-too-human place. “People always ask the question of, ‘Is the AI coming for our jobs?'” she says. “It’s not the AI that’s coming for our jobs. It’s the people that are wielding the AI.”

Plus, some AI-made music can even be a lot of fun. Dr. Palamara personally enjoys some music created by AI, citing a Ray Charles song that’s been mixed with a Nickelback track, and a version of Johnny Cash singing “Barbie Girl” in the style of “Folsom Prison Blues.”

AI is going to change our world one way or another, so it’s critical to focus on shaping it into something we actually want to see in the world. As Evans explains, “Artists have been threatened by new technologies since the beginning of time.” She wants to urge artists to try to embrace AI as a tool, just like that fancy new pedal or recording software.

As she puts it: “I think if you look at the history, the most effective way for artists to combat displacement or exploitation is to find a way to take the threatening new thing and make it part of who they are.”