My book, Build or Die: How America Suffocates Its Cities and What to Do About It, will be published by Princeton University Press on December 8, 2026. Preorder the book now from Bookshop.org.
In “Pierre Menard, Author of the Quixote,” a short story by Jorge Luis Borges, the titular character, a 20th-century French poet and essayist, sets out to “compose the Quixote.” Not a translation of Don Quixote, much less a contemporary spin on the original; in the words of the story’s narrator, Menard goes through a series of mental and physical exercises that will make it possible for him to "produce a number of pages which coincided—word for word and line for line—with those of Miguel de Cervantes."
The story’s humor and central idea come from the narrator’s insistence that Menard has written a better Quixote than Cervantes, even though the two texts are literally identical. The only difference between them is the context in which they were written. For example: "Cervantes crudely juxtaposes the humble provincial reality of his country against the fantasies of the romance, while Menard chooses as his 'reality' the land of Carmen during the century that saw the Battle of Lepanto and the plays of Lope de Vega."
The story is essentially an extended parody of modern literary criticism, or at least one strain of literary criticism. And it’s a pretty funny parody. But when I read the story now, I can’t help but think Borges’s unnamed narrator has a point. The context in which an artwork is created and the means of its creation are both part of the final product. When we read a novel or poem, or listen to music, or watch a movie, part of what we’re doing is absorbing the hyper-concentrated accretion of an artist (or group of artists’) life experiences, choices, and artistic practice. That’s what makes good art so personal and idiosyncratic, and it is also why there is some truth to the cliché that all writing is a form of autobiography.
I’ve been thinking about what goes into artistic creation ever since I learned of the recent death of jazz giant Sonny Rollins at the age of 95. Rollins was one of the greatest saxophone players of all time; some of this greatness might have come from his native genius, but a lot of it had to do with the fact that he drilled himself on the saxophone relentlessly. Famously, between 1959 and 1961, Rollins spent up to 16 hours a day practicing his instrument alone on the Williamsburg Bridge. He also developed a disciplined spiritual practice built around yoga.
I think you can hear the fruits of both these disciplines in his music, and not just because he was extremely technically proficient. This is a man who really knew his saxophone and was dedicated to exploring the full range of its rhythmic and melodic possibilities. Like any good jazz practitioner, he was also clearly interested in timbre, and subtle variations in sound and color that are untranscribable in Western musical notation.
In other words, Rollins was a consummate artist. His life is, in fact, an object lesson in what it means to be a true artist. Compare his obsessive self-discipline with the sort of just-one-click artistic output promised by generative AI’s boosters. If you ever pop over to X (which, to be clear, I don’t recommend), you’re likely to encounter some venture capital guy talking about how generative AI democratizes art by allowing people to “draw” without learning how to draw, or “compose music” without knowing a single thing about the art of composition.
But that’s not really artistic creation, no matter how convincing the results may be. Because the point of creation is never just the end product; it’s the practice. Rollins didn’t spend 16 hours a day on the Williamsburg Bridge trying to achieve a particular outcome. He was exploring musical possibilities and changing himself in ways that even he may not have understood. That process of exploration and self-definition is itself the point. It’s also the reason why his music conveys something meaningful, something irreducible to a verbal synopsis.
It’s possible to imagine a generative AI program that can perfectly reproduce a Sonny Rollins track that isn’t in its training data. We’re already years past the point where such programs can create “new” Bach compositions. But the narrator of “Pierre Menard” would pan these compositions, and he’d be right to do so.
That isn’t to say you can’t make art with generative AI. Another favorite musician of mine, Brian Eno, was developing procedurally generated soundscapes well before the current AI boom. But the process by which he arrived at these soundscapes is extremely important. He didn’t just type “give me a very long track that sounds like Brian Eno” into a chat box. In the case of his 2017 piece Reflection, a procedural algorithm that endlessly generates new musical variations, the app itself is part of the work of art. It is the result of a theoretically infinite number of subjective taste judgments and micro-adjustments made by Eno and his collaborators. It is also the product of one man’s more than 50 years of restless experimentation with ambient music. Eno, the artist, permeates Reflection; in a sense, he is offering an example of how generative models can be part of an artistic practice instead of serving as a hollow substitute.
Those years of work, that development of an idiosyncratic taste and sensibility, and all of those small choices are where art comes from. It originates somewhere in the hazy territory between mastery and surrender: mastery of one’s craft, but also surrender to one’s own instincts about what “feels” right. The generative models being pushed by those aforementioned X denizens minimize the role of both personal mastery and personal taste; instead of democratizing the ability to make art, they encourage people to fake artistry without engaging in the set of practices that gives art its meaning.
From one perspective, “Pierre Menard” is a sort of prank on literary critics, but from another, there’s something very admirable about its title character: he understands that to produce Don Quixote, he first needs to undergo a set of personal transformations. Borges makes clear that he isn’t trying to become Cervantes: he is trying to become the version of Pierre Menard who would write Don Quixote, a subtle but rich distinction.
Sonny Rollins was engaged in a purer artistic practice because he wasn’t trying to reproduce anything; he was exploring the possibilities of his own sound so that he could create Sonny Rollins’s music. There’s a lesson there, and not just for full-time artists.