AI Goes to Court
...with a near-future on the line.
In recent months, AI audio companies like Suno and Udio have been in the news for the promise of their text-to-tunes tech. Just type in a few phrases, and… an original piece of music of your very own, created in seconds. It’s a revolution! After all, who wouldn’t want…a “dancehall” song about two podcasters talking about music and capitalism? It’s the future we were promised!
At least that’s the narrative being pushed by the world of venture capital, which has thrown real money at the fledgling firms— 125 million and counting for Suno, and at least another 10 for Udio. These ideologies are echoed by the companies themselves, which claim that their services (currently, although inevitably not permanently, extremely cheap) are designed to address a world in which:
“Somewhere along the road, we lost some of the joy of making music. A few of us grow up to be Real Musicians, with the requisite training or innate skill, while the rest of us come to understand that our place is to sit quietly in the audience and enjoy the show… We started Suno to build a future where anyone can make music.”
(tears. gratitude. a Pepsi stopping the riots.)
But of course, even the flashiest of tech exists in a world that has both a past (i.e. over a hundred years of for-profit music making) and a present (an entire industry still devoted to doing the same). To try and understand what these companies are actually promising—and what they could do to the music industry—we begin our most recent episode by thinking through some possible futures, from the mind-bendingly good to the vast universe of echoing slop.
But of course, even if the tech is there (and the machine learning models learn what dancehall is), the world might not follow along as smoothly as the CEOs would like. In particular, the major labels, incensed at what they believe was the wholesale theft of their copyrights, have launched a series of lawsuits aimed at kneecapping the wannabe unicorns.
This month, Suno and Udio responded in startling fashion. Yes, it turns out, they did indeed train their models on recorded music. But it wasn’t stealing, because the recordings were already online? They’re vociferously claiming that this type of industrial-scale scraping falls under fair use. And…they might be right. Not because it actually makes much sense—certainly, the argument that permanently structuring the weights of a massive machine-learning model is the same as a kid watching blues videos on Youtube feels…a bit forced. But rather, because these types of court cases are almost inevitably more about mediating between the various economic powers seeking to earn profit from a given space.
Without a doubt, the emergence of AI music is going to require a serious rethink of copyright. But a look to past moments of extreme technological flux in this space (think, for example, about the emergence of recording in the early 20th century) have rarely, if ever, produced common-sense results. (A personal favorite of that earlier moment was the spectacle of befuddled congressmen peering closely at wax cylinders, seeing if they could “read” the music contained in the grooves.”)
Given the amount of money that has poured into this part of the tech sector—coupled with growing concerns that the productivity gains promised by OpenAI+ Co are simply not there—it’s safe to expect a LOT of attention on this type of potentially precedent-setting conflict. At stake is more than just the future of not having to learn Garageband in order to make mediocre house. Instead, the battle over audio is shaping up to be a defining moment for generative AI more generally—a conflict with billions on the line.
But back to the lawsuits…as we noted in our show, the big labels are staring the potentials of generative AI down with clear eyes. This isn’t Napster all over again. The technology is here and the label heads know that. But it might be remiss to view this as just them wanting a slice of the pie (AKA another licensing deal). Instead, it’s worth speculating (at least!) that unlike UMG’s recent battle with TikTok, the lawsuits being brought against these AI companies by the big labels already reveal something more interesting: that what these companies like Udio and Suno are doing is exactly what Grainge and Co. want to do themselves.
Hang with us a bit on this. In the last third of our episode, we talk through potential responses to AI within different creative industries. In particular, we focused on the ways that the basic nature of music as a commodity is tightly connected to the technology of the day (vinyl to CD to MP3 to streaming). This is why these changes become such an existential question for the music industry. They don’t just disrupt a profit flow. They disrupt the basic nature of commoditization—the ground on which the industry sits, so to speak.
But it’s worth also speculating that this technology offers something different: the ability to create music without the need for artists, musicians, producers, etc. And it can all be done in-house. Now, try to say that this wouldn’t be attractive to Lucien Grainge: a fully integrated production house within UMG, one that operates sans artists and all the scandals, contract, problems and paychecks that goes with them. At best, the technology generates pop hits and avatar stars and the big label runs and has total control over the IP underlying all of it. At the lowest end, the kind of passive muzak being played everywhere from malls to gyms to gas stations is created by this technology and the majors (again) own all of it and collect 100% of the bag. Not so bad.
This isn’t to suggest a future where there are no more human pop stars and/or you’ll stop hearing Todd Rundgren at CVS. But if a large part of the current industry could be sent a pink slip and made automated and entirely in-house; what in the name of all that is Capital would stop the majors? Possibly, a union! (See: Hollywood and recent strikes). Oh right, there really is no music union. Welp.
It’s all speculation, baby. But don’t be shocked if what Grainge and Co. are fighting today becomes what they’re pushing for in some not-so-near future.
You can listen to that hear (Itunes), hear (Spotify), or hear (Podbean)
Department of Actual Music:
Sam: Every so often, music writers have a really stupid argument about “what the greatest american band” is. It’s Chic. It’s always been Chic. Over the last few years, I’ve returned again and again to the incredible music contained on the group’s 12-inch singles. As propulsive and minimal as anything produced by Neu!, as funky as James Brown, as catchy as Motown. And every time I listen, the groove only seems to deepen. Incredible.
Saxon: Obscure London band from the late 60s featuring a singer with one lung. Apparently, the boys never got royalties for this record (or subsequent re-releases). A shame. Poor man’s garage Zeppelin. Yes, plz.
Saxon and Sam