AI Is Sure to Transform the Music Biz. Opinions Vary on Exactly How.

AI was an omnipresent topic at the Music Tectonics conference in Santa Monica, Calif., earlier this week, creeping into seemingly every panel discussion and casual poolside conversation. Everybody can see that AI will transform the music business. That’s a 30,000-foot view. Zoom in, however, and there’s far less certainty about how, exactly, AI will disrupt the status quo. 

“Not everybody wants to be a creator” was a frequently heard sentiment. Lucas Cantor Santiago of Mindset Ventures has a particular point of view as a composer. The setup he currently uses to write music would have cost $200,000 15 years ago. Now, somebody can get “basically the same tools” from a trip to the Apple Store. But Cantor Santiago doesn’t believe access to tools has led to more creators. “It’s just caused people like me to start writing music faster, and maybe people who didn’t have classical training to be able to start writing music,” he said on a panel.

AI’s ability to help human creators, not replace them, was a common theme at Music Tectonics. Granted, the conference was heavy on consumer technology brands such as Yamaha, Roland and Fender. Had AI anarchists been invited to speak, there would have been more diversity of thought. But the opinions of people who actually make music for a living carry a lot of weight, as they’re on the front lines of making music that eventually finds an audience. To this crowd, AI slop has little redeeming value and won’t find a meaningful audience. 

The historical record doesn’t fully support the idea that AI won’t increase the ranks of creators, though. Greater access to inexpensive production and distribution tools has already transformed the music business. Artists who were previously locked out of nationwide distribution — it was impossible for a DIY artist to get Tower Records to stock their CDs — now have access to tens of millions of consumers through digital distributors and digital service providers (DSPs) such as iTunes and Spotify. “When I owned a record label and house label in Chicago in the ‘80s, there were 100 new records a week,” said Matt Adell, co-founder and COO of Musical AI, on a panel. “When I left [EDM download store] Beatport, there were 27,000 new records a day. There are now over 150,000 new songs a day hitting the DSPs.”  

Given easier tools, people are already creating more music. Many of the 150,000 songs a day cited by Adell — or whatever the number is currently — were created by AI. French music streamer Deezer said in September that 28% of tracks uploaded are created wholly by generative AI, underscoring the fact that AI tools could lead to more music being created. Unpopular, long tail music may not attract much attention, but it creates markets where none previously existed. DIY distributors such as DistroKid, CD Baby and TuneCore can operate because production tools are inexpensive — sometimes free — and artists can afford the modest fees to distribute their songs globally.  

AI’s biggest impact could be to turn everybody into a small-scale creator. Kristen Bender, senior vp of digital innovation strategy and business development at Universal Music Group (UMG), noted during a panel that 30% to 40% of all music content on social media platforms has been manipulated by AI in some way, suggesting there are more creators than people might think. “We think that AI is going to enable so [much] hyper-personalization and interesting ways to interact with content,” she said. 

Along those lines, Liz Moody, a partner at law firm Granderson Des Rochers, described how AI tools will allow fans to interact with artists in new ways. Moody, who worked on Udio’s recent licensing deal with UMG, told the audience Udio could create “a fan-focused experience where fans can work with their favorite artists to make personalized music, maybe with the artist’s voice, or maybe create some mashups between two songs that they love.” 

When AI tools first appeared, the initial conversation focused on AI-generated music’s potential to supplant the popularity of human-created songs. But Bender and Moody — who have visibility into where these business models are headed — encouraged people to think smaller. It’s easy to imagine a licensed, industry-sanctioned generative AI platform partnering with well-known artists to create personalized renditions of “Happy Birthday” for their subscribers. But it’s a lot harder to imagine anyone other than the creator wanting to hear their personalized version.