The Greatest American Hero was killed by a machine. We all know the story of John Henry, the mythical hammer man who competed against a steam drill to tunnel through a mountain and beat the machine—only to die in the process. Now with the rise of AI, a new machine is competing against humans, only this time it’s battling those who make those myths and stories: artists, writers, and musicians.
A few months ago, Vassar creative writing professor David Means was contemplating this struggle, which had been consuming many of his conversations with students in writing classes. One afternoon, while wrestling with a strange story he was writing about a cat, a thought occurred to him: “AI can’t write this story I’m working on now because I haven’t finished dreaming it.” The story, he realized, was still locked inside his head, and would never be freed by a computer despite a neural network’s best attempt to ape his own neurons. Taking a break from his battle with the story, he wrote out his thoughts in what became an essay for the New York Times, “AI Can’t Write My Cat Story Because It Hasn’t Felt What I Feel.”
“Art is a deep form of expression—human expression—and part of our experience of it has to do with the sense that at the other end of the process is a person who created the work, dreamed it up, or made specific choices,” says Means about his epiphany that solitary afternoon. We admire Edward Hopper, he says, not just for his technical skill in painting, but for his unique eye in seeking out isolated individuals and lonely-looking buildings and using them to say something about the loneliness of American life in the 20th century. “Is AI going to wander lonely streets and feel sad and live inside the complexity of relationships and then create art?” Means asks.
At best, he wrote in his essay, AI can mimic artwork, or help a human create art—but it could never replace a writer like himself because “AI has never felt what I’ve felt” or moved through “the emotional matrix of living a singular, individual life.” In fact, he argues, true artwork is original and disappears once the story is on the page or painted on the canvas. “Originality lies in the conception before the work is published or exhibited,” Means wrote. The art “is simply the final product of a complicated, often illogical process and, in some ways, is the least important aspect of the endeavor.”
That goes for pop songs as well. The music industry was shaken this spring when an anonymous creator released a track called “Heart on My Sleeve,” impersonating the musical style of Drake and the Weekend—leading wags to call it “Deepfake Drake.” But whether or not we can tell the difference between a real and AI-generated song is beside the point, says Means: There’s something about knowing the human experience behind the writing of the song that helps it cross the void between people. “We love hearing the stories behind the songs we hear,” Means says. “Think of Taylor Swift or Kendrick Lamar. Listeners have a relationship with the music and the artist. I’m not sure how AI is going to go on tour or do readings at bookstores.”
Even as music, AI-generated songs seem to fall flat, noted Vassar Associate Professor of Philosophy Barry Lam in a recent episode of his philosophy and pop-culture podcast “Hi-Phi Nation” focusing on AI-generated music. Deepfake Drake falls under the category of something called vocal emulation, Lam says, using AI to mimic someone else’s voice when singing a human-generated composition. “I’m not too worried about that,” he says. “The history of the music industry shows it’s so litigious, I have no doubt it will find a way to sue that out of existence, or monetize it in a way that’s satisfactory.”
What about original compositions by AI, though? That too is a long way from threatening human creativity, Lam says. “Fully generative stuff just sounds glitchy and weird,” he says. The best machines can do now is to create a composition based on the work of another artist, translating it into notes on the page that can then be scored and sung by humans (or a vocal emulator). “If you said, ‘Here’s the complete oeuvre of the Beatles—generate a new John Lennon song,’ it’s going to create something that’s derivatively like John Lennon,” says Lam. “[Or] at least as good as the Monkees—which is not bad!”
In the podcast, Lam explores Google Music’s latest efforts to create original music based on the style of certain genres, such as club music from the 1970s or electronic music from the 2000s. Even when it creates original songs, they sound derivative of music that already exists, he says, essentially creating inoffensive background music to study or exercise to—or filler music for a YouTube video or podcast when someone can’t afford a human composer. This may be putting some people out of work, to be sure, but it’s not creating the mass creative crisis some technophobes fear.
“At the same time, it’s going to unlock a lot of people being able to create other things—like film students having royalty-free music for film scoring,” he says. “It’s a weird economic trade-off.”
As far as creating a new pop hit, however, Lam sees that capability as being a long way off. In fact, he says, the only time it really creates something interesting is “when it strays from sounding like anything familiar”—for example, when Google asks AI to create a musical composition based on Van Gogh’s Starry Night or Edvard Munch’s The Scream. By training AI on humans’ subjective descriptions of different kinds of creative art, Lam says, AI could help us conceive of a universal language of creativity, translating visual art to musical art or vice versa. “Maybe you could take the best meal you’ve ever had, and say here is an opera based on it.”
As far as replacing artists, writers, and musicians, however, AI seems to be losing the battle. Means wonders if it might actually help human artists by giving them a new currency for authenticity in the age of mechanization. “When the Industrial Age began and we could suddenly replicate things—furniture, for example—people began to long for antiques, for the real things,” Means says. Isn’t that, after all, the point of the John Henry story? That even in death, his spirit triumphed over the machine that tried to take his dignity, to the point where we remember his name today. If trends continue, says Means, “I’m sure that artists are going to put a stamp on their product that says: Written by a human with a life.”