When the internet revolution arrived, music was the canary in the coal mine. It was the first industry to choke on the digital carbon monoxide. As soon as masterpieces such as Dark Side of the Moon and Thriller were reduced to binary code, there was trouble inside the mine.
The crisis caught the industry flatfooted. Labels balked at modernizing (“CDs today, CDs tomorrow, CDs forever!”) Once the magnitude of that mistake was apparent, they attempted a rear-guard defense with litigation. It failed.
The most famous example was Metallica’s crusade against Napster. The band’s drummer, Lars Ulrich, went before a Congressional committee to explain that file-sharing was “a sophisticated version of taping a baseball game off the television”, tantamount to “theft”. For that blunt assessment, The New York Times pilloried him with the headline: “Metallica Turns On Its Fans.” Ulrich and his band were portrayed as fat-cats in studded leather, eager to exact economic pain on their working-class listeners.
The economic pain was instead doled out to working musicians. Royalties dried up, album budgets shrank. No more mailbox money, baby.
It’s no surprise then that many musicians looked on with bitter glee as the same digital revolution consumed the news media. After all, it was the same media that had vilified them for raising the alarm. You don’t think Chicken Little harbored a dark little smile as the sky fell upon his ridiculers? The idea of a fully-staffed local newspaper is now just as anachronistic as a fully-stocked record store.
(In a supreme twist of irony, the only newspaper to fully survive the digital onslaught did so by embracing what Lars Ulrich had been advocating for: paywalling. The New York Times dug their digital moat in 2011 and now remains financially garrisoned because of it. Well… that and the acquisition of ‘Wordle’.)
Musicians eventually dug a moat of their own. Digital streaming. Spotify spearheaded a new model where consumers paid for music again. And while many good faith criticisms can be leveled at the Green Giant, the years between Napster and the modern streaming economy were indeed bleaker. So credit where credit is due.
The lesson is clear, though. When a technological revolution is nigh, it’s best to adapt in anticipation, not in reaction. To do so is to dictate the terms of a victory; rather than to surrender to the terms of a defeat. Music’s adaptation to internet file-sharing was reactive. That is to say, it was too late. By the time streaming arrived, recorded music had already been devalued. It was no longer fine wine to be paid for, it was tap water to freely chugged from the spigot. The meek terms of a surrender.
The Artificial Intelligence revolution is now here. Music is once again the coughing canary. Consumer-facing AI engines can now create entire songs from simple language prompts. On the current course, streaming services will soon be flooded with music that is wholly AI generated. The moat of digital streaming itself will be catastrophically diluted.
Musicians need not despair, though. This time, the industry appears to be adapting with alacrity. The first shot across the bow is a landmark lawsuit spearheaded by the Recording Industry Association of America (RIAA) against two of the leaders in generative AI music: Udio and Suno. For the sake of clarity, let us focus solely on Suno.
Whereas other music generation engines can discreetly create instrumentals, lyrics or voices, Suno can do all three combined. It does so in an instant. The time it takes for an artist to compose and record a song varies widely; but it certainly takes longer than typing the prompt: “shoegaze pop song in the key of B major at 120BPM with lyrics about my dead dog”.
Simply put, humans cannot compete. In the time it takes an artist to compose and record a single song, AI can generate a tsunami of competing recordings. As an artist sleeps, the unblinking eye of Suno will compose whole sonic libraries.
The lawsuit hinges on how those companies ‘trained’ their AI algorithms. To create ‘new’ material, the AI engines had to ingest an immense amount of previously recorded music. In the same way that ChatGPT scraped all the data from Facebook and Wikipedia to train its learned language models, Suno scraped the entirety of recorded music history.
But here’s the rub. Recorded music enjoys legal protections. It’s not defenseless like your Aunt Sheila’s Facebook post about the Illuminati or a Wikipedia entry on Guy Fieri’s favorite fried rice. Recorded music is copyrighted. This lawsuit alleges that Suno has committed copyright infringement at an “almost unimaginable scale”.
At first blush, it does seem to me like copyright infringement. But that conclusion leads to an uncomfortable question: When I was learning to write songs, what did I train myself on? Was I not, in fact, involved in the same appropriation as Suno?
When I arrived in Chicago at the age of 22, I barely knew a soul. I was a stranger in a strange land; quite literally in Oz. I walked like a ghost between the shoppers on Michigan avenue. I passed invisible on bridges above the tourists gliding down the Chicago River on architectural boat tours- teachers from Indiana and Ohio marveling at the towering brick monuments built by their forefathers.
My loneliness and boredom afforded me time to consume art that I loved. It wasn’t long before I began trying to make my own. So I would listen to Bob Dylan’s The Freewheeling and learn his fingerpicking patterns. Or I would leaf through Walt Whitman’s Leaves of Grass and study his rhetorical devices. Or maybe, I’d listen to The Band’s Music from Big Pink and find a compelling chord progression. Add a Dylan picking pattern and a Whitman rhetorical flourish on top, and voila! A song of my own. It was the thinking man’s paint-by-numbers.
Listening to my debut album, it is easy to hear those influences. Maybe too easy. I didn’t just wear them on my sleeve, I got them sleeve-tattooed!
The artists that I was borrowing from, though, were not laboring in a vacuum themselves. Dylan’s first albums were homages to his hero Woody Guthrie. And upon reading Whitman, Ralph Waldo Emerson would have recognized his own rhetorical devices marbled throughout.
“Good artists borrow, great artists steal” is not just a maxim, it is an injunction. It is a demand to artists: to make everything old, new again. But that injunction is the wellspring of countless tough questions:
Won’t the creators of AI music engines simply argue they are following in this same tradition?
Are they simply doing with Code what I was doing with the left hemisphere of my brain?
Does opposing their progress just make me a medieval monk railing against the Guttenburg Press? Or stable-owner bemoaning Henry Ford’s horseless carriages?
If something is objectively beautiful, does it matter how it is made?
These questions, though, quickly take on the crude sophistry of a dorm room debate; elaborate exercises in missing the whole damned point. Sure, an undergrad can stay up until midnight smoking cigarettes and debating ‘free will’. But a decade later, if that same undergrad runs a Ponzi scheme or cheats on his wife, no one wants to hear the word ‘determinism’. This is real life, man.
So it goes with generative AI. What makes a human songwriter’s appropriation of prior work legitimate, and a supercomputer’s illegitimate? The fact that we prefer it to be so. That’s enough. No further justification needed. Creativity is a bedrock human endeavor. We can decline to participate in its outsourcing.
It is edifying to spend ten thousand hours playing the guitar. It is good for the soul to sit in front of a blank page for an afternoon. It is beautiful to learn from an older songwriter how to write a middle-eight. Preferring those modalities does not make a person a luddite.
One of the greatest lies of modernity is that the advance of technology is inexorable. That we, as humans, are tiny pebbles being swept along by its irresistible current. Nonsense. We have agency. Always and forever, humans have forbidden certain categories of behavior. Our toolkit for doing so has been honed over centuries: taboos, taxes, boycotts, sanctions, subsidy removal. And yes, lawsuits from the RIAA to protect intellectual property.
I am not suggesting that we go full-Amish on AI. We should not entirely foreswear it. It should just be employed to the degree that it promotes human flourishing, and curtailed to the degree that it does not. As the author Joanna Maciejewska put it: “I want AI to do my laundry and dishes so that I can do art and write, not for AI to do my art and writing so that I can do my laundry and dishes.”
ChatGPT instantly delivering a bibliography of historical interpretations of Dostoyevsky? Sure.
Writing my essay on the meaning of “Crime and Punishment”? Nope!
Midjourney automating the loops in a vast pattern that is part of a design? Sounds good.
Designing my city’s Cathedral? That’s a ‘No’ for me, dawg.
As a recovering ironic hipster, it is agonizing to admit that Lars Ulrich was right. Metallica was completely justified in their lawsuit. We should not have let an entire generation get accustomed to not paying for music. And now, that camel’s nose cannot be forced back out of the tent.
Faced with the generative AI revolution, artists still have a choice. We can maintain our hip defeatist attitude toward technology- “I guess they’re going to screw us again!”. We can dismiss the Metallicas of the world pulling the alarm- “Look at those rich chuds just trying to buy their third beach house”. But if we impotently sneer our way through this revolution, there will be no turning back. Human-made music will become like the majestic brickwork that tourists flock to see in Chicago: a majestic art the dead once knew but which the living have forgotten.