is a reporter who writes about tech, money, and human behavior. She joined The Verge in 2014 as science editor. Previously, she was a reporter at Bloomberg.
Sure, everyone hates record labels — but the AI industry has figured out how to make them look like heroes. So that’s at least one very impressive accomplishment for AI.
AI is cutting a swath across a number of creative industries — with AI-generated book covers, the Chicago Sun-Times publishing an AI-generated list of books that don’t exist, and AI-generated stories at CNET under real authors’ bylines. The music industry is no exception. But while many of these fields are mired in questions about whether AI models are illegally trained on pirated data, the music industry is coming at the issue from a position of unusual strength: the benefits of years of case law backing copyright protections, a regimented licensing system, and a handful of powerful companies that control the industry. Record labels have chosen to fight several AI companies on copyright law, and they have a strong hand to play.
Historically, whatever the tech industry inflicts on the music industry will eventually happen to every other creative industry, too. If that’s true here, then all the AI companies that ganked copyrighted material are in a lot of trouble.
Can home prompting kill music careers?
There are some positive things AI music startups can accomplish — like reducing barriers for musicians to record themselves. Take the artist D4vd, who recorded his breakout hit “Romantic Homicide” in his sister’s closet using BandLab, an app for making music without a studio that includes some AI features. (D4vd began creating music to soundtrack his Fortnite YouTube montages without getting a copyright strike for using existing work.) The point of BandLab is giving more musicians around the world the opportunity to record music, send it into the world, and maybe get paid for their work, says Kuok Meng Ru, the CEO of the app’s parent company. AI tools can supercharge that, he says.
That use, however, isn’t exactly what big-time AI companies like Suno and Udio have in mind. Suno declined to comment for this story. Udio did not respond to a request for comment.
Suno and Udio are designed to let music consumers generate new songs with a few words. Users type in, say, “Prompt: bossa nova song using a wide range of percussion and a horn section about a cat, active, energetic, uptempo, chaotic” and get a song, wholesale, without even writing their own lyrics. The idea that most listeners will do this regularly seems unlikely — making music is more work than just listening to it, even with text prompts — as does the idea that AI will replace people’s favorite human artists. (Also, the music is pretty bad.)
“AI flooded the market with it.”
A lot of listening is passive consumption, like a person putting on a playlist while doing the dishes or studying, or a business piping background tunes to customers. That background music is up for grabs — not by consumers, but by spammers using these tools. They’re already generating consumer-facing slop and putting it on Spotify, effectively crowding out real artists.
... continue reading