What Becomes of Creativity in the Age of Artificial Intelligence
There is a strange kind of frugality in real art. Not the frugality of money, but of energy — of time, attention, of lifespan. Art is the economical curation of all possible expressions of an idea or a feeling. It is not abundance that makes it human, but limitation.
An artist searches for the ideal form of something that moves inside them. That search is slow, costly, and finite. Each brushstroke, each failure, each sketch is a calculation: what still deserves energy, what no longer does. The studies, the variations, the detours — all of them are steps in a ritual of selection. In the end, one image remains that gathers everything: the idea, the emotion, and the loss along the way.
Artificial intelligence knows nothing of this economy. It burns no lifetime, it wastes nothing of itself. AI can generate thousands of variations without dying from a single one of them. And that is precisely what makes it suspicious: it lacks the necessity of choice. For humans, choosing is survival. For the machine, choosing is computation.
But perhaps that framing is too simple. The machine does operate under constraints — just different ones. It inherits the biases and boundaries of its training data, the architectural decisions of its designers, the feedback loops of human preference. These are not the constraints of mortality, but they are constraints nonetheless. An AI cannot paint what it has never seen encoded in its weights. It cannot write beyond the grammar of its corpus. Its abundance is not infinite; it is large but circumscribed, like a vast library with invisible walls.
The question, then, is not whether the machine experiences limitation, but whether its limitations mean anything to it. Does a constraint shape the machine’s choices the way scarcity shapes ours? Or does it simply redirect the calculation without changing its nature?
For the human artist, constraint is felt. It has texture. The painter knows when the hand tires, when the eye loses focus, when an idea has exhausted its claim on attention. These sensations are not incidental to the work — they are information, signals that guide judgment. The machine receives no such signals. Its constraints are external rules, not internal experiences. It does not feel the weight of its own limitations.
And yet, that is also its promise. As a tool, AI can widen the horizon — enlarge the field of what is possible. It can bring the artist closer, faster, to that one image — not by replacing them, but by illuminating the field in which they search. The machine opens the paths; the human decides where the light stays. Though in practice, the process is messier than this binary suggests — a dialogue of prompts and responses, selections and refinements, where agency is distributed rather than cleanly divided.
But what happens when the machine learns to predict not just forms, but preferences? What happens when AI becomes so attuned to what moves people, what survives cultural memory, what accrues value over time, that it can curate better than we can? If curation itself becomes a pattern the machine can learn, then the boundary between human choice and algorithmic recommendation begins to dissolve.
We already see this happening. We allow algorithms to choose our music, our news, our next purchase. We trust them to surface what we didn’t know we wanted. If an AI can learn what art resonates across generations, what stories lodge themselves in collective memory, what images refuse to fade — then what remains distinctly human in the act of choosing?
Perhaps the answer lies not in the act of choosing itself, but in the stakes of the choice. When a human selects one image over another, they do so knowing that time spent here is time not spent elsewhere. The choice carries the weight of finitude. When an algorithm curates, it optimizes, but it does not sacrifice. It does not feel the loss of the unchosen paths.
This is not a small difference. It is the difference between curation as calculation and curation as commitment. The human artist does not merely select — they stand behind the selection. They offer their work as a kind of testimony: this mattered enough to spend myself on it. The machine cannot testify in this way. It has no self to spend.
But there is another dimension to this debate, one that runs beneath the philosophical questions like a hidden current: the question of access, of class, of who gets to make art at all.
Before AI, before cameras, before mechanical reproduction, art was sorted into hierarchies — highbrow and lowbrow, Rembrandt and the weeping child. These classifications pretended to be about quality, about aesthetic merit, about what was “real” art. But often they were simply marking territory along class lines. A Rembrandt hung in the homes of those who could afford one. For the baker, the tailor, the shopkeeper, a mass-produced print of a weeping child was the best they could do. It was what beauty looked like when you couldn’t pay for transcendence.
The anxiety about AI-generated art echoes these old debates. Perhaps it is the fate of AI art to be classified as lowbrow — the digital equivalent of the weeping child, abundant and cheap, accessible to anyone with a prompt and an internet connection. And perhaps the resistance to it is not only about authenticity or effort, but about gatekeeping. About who gets to call themselves an artist.
Every technological shift in art has faced similar resistance. The camera was accused of killing painting. The printing press was accused of debasing literature. Synthesizers were accused of making “fake” music. Each time, the boundaries of what counted as art had to be renegotiated. (I have written about this elsewhere on khadims.blog, and others have made similar arguments, so I will not belabor the point here.)
What history shows is that the tools do not determine the outcome. Photography did not kill painting — it freed painting to become something else. The printing press did not debase literature — it created new literatures, new audiences, new forms. And from the mass accessibility of these technologies, new masters emerged. Photographers who made the camera sing. Writers who understood what print could do that manuscript could not.
The same will happen with AI. There will be noise, abundance, a glut of generated images indistinguishable from one another. But there will also be artists who do something new with the tool — something that could not have been done before, something that makes us stop and reconsider what the medium can hold. The AI Rembrandts have not yet appeared, but they will. And when they do, we will recognize them not by the tool they used, but by what they chose to make of the abundance.
And so we return to the question: are we still artists, now that AI paints, writes, and thinks beside us?
Yes — as long as we remain the ones who choose what endures. Not because we choose more accurately than the machine, but because our choices carry a different kind of authority. They are underwritten by mortality. They are signed with the hours we cannot recover.
What will endure, we cannot say in advance. Art resists definition. Every time we draw a line around what counts, some future artist crosses it, and we discover the line was arbitrary all along. But the act of choosing — of committing to the belief that this deserves to last, even though we cannot predict whether it will — that act remains human. It is not a calculation. It is a wager, a bet that this, of all things, is worth the piece of life it cost.
The machine can help us search. It can multiply possibilities, reveal patterns we might have missed, accelerate the process of exploration. But it cannot replace the moment when the artist says: here, and no further. That moment is not an optimization. It is a declaration of value in the face of uncertainty.
As long as we still make that wager, we are still artists. The machine may learn to predict what endures, but only we can decide what is worth the spending.
We already have a model for how this works. Consider memes — perhaps the most democratic art form of our time. Every day, thousands of images, videos, and phrases compete for attention. Algorithms help distribute them, surface them, amplify them. But the algorithms do not decide what becomes iconic. Humans do that. Humans decide what’s funny, what’s cutting, what captures a moment so perfectly it enters the collective vocabulary. No AI could have invented “loss.jpg” or understood why “distracted boyfriend” resonated across cultures. Those required context, timing, a feel for the collective mood. The algorithm just helped the good ones spread faster.
Replace “memes” with “art” and you have your answer. The machine can accelerate distribution, multiply possibilities, help surface what moves people. But it cannot replace the human judgment that says: this one matters. That judgment is not infallible — it is messy, contested, shaped by fashion and power and accident. But it is ours. And it operates at a scale and speed that already involves machines without being reducible to them.
Perhaps, in time, we will learn not to fight the abundance — not to be Quixote charging at windmills — but to navigate it, to find within the infinite what still deserves our finite attention. The question is not whether AI will make art. The question is whether we will let it teach us new ways to choose, new ways to curate, new ways to commit our limited lifespans to what we believe should outlast us.
Khadim Zaman's Blog 