Why Nobody Can Agree on What Art Is: 2,500 Years of Arguments

In 1917, a porcelain urinal signed "R. Mutt" was rejected from an exhibition that had promised to show anything — and the argument over what counts as art has never really recovered. But the urinal didn't start the fight. It was round four of an argument that had already been running for more than two thousand years. This is the story of that argument: who landed which blow, why no definition has ever survived contact with actual artists, and why the same fight is now raging again over AI. For the definitions laid out side by side, see Part 1 of this encyclopedia — this companion piece is about why they keep colliding.

Round One: Plato Tried to Banish the Artists

The first surviving definition of art is an attack on it. In Book X of the Republic (c. 380 BCE), Plato argues that a painting is a copy of a copy: a painted bed imitates a carpenter's bed, which itself imitates the ideal Form of a bed — so art sits "three removes from the truth." Worse, it feeds the emotional, irrational part of the soul. His conclusion was not a definition but a sentence: most poets and painters should be kept out of the well-run city.

His student Aristotle quietly landed the counter-punch in the Poetics. Imitation, he argued, is how humans learn — children do it instinctively — and tragedy purges and clarifies emotion rather than merely inflaming it. Art wasn't a defective copy of reality; it was a way of understanding it. Notice what just happened, because it happens in every round that follows: the definition each thinker produced perfectly fit — and defended — the art they already valued.

Round Two: Skill, Genius, and the Academy

For most of the centuries after Aristotle, the argument went quiet because the answer seemed obvious: ars — the Latin root of "art" — meant skilled making, and a medieval goldsmith and a painter belonged to the same world of guilds and workshops. Two developments restarted the fight. In 1550, Giorgio Vasari's Lives of the Artists recast painters and sculptors as inspired geniuses rather than tradesmen. Then the academies of the 17th and 18th centuries tried to settle the question by rulebook: correct subjects, correct technique, correct taste — with juries to enforce all three.

That settlement held until the 19th century, when photography took over art's documentary job and painters began asking what painting was for, if not accurate depiction. The stage was set for the loudest round of all.

Round Three: Tolstoy Says It's About Feeling — and Condemns Almost Everything

Leo Tolstoy's What Is Art? (1898) is the most radical definition ever offered by a great artist. Art, Tolstoy wrote, is the transmission of feeling: one person, "by means of certain external signs, consciously transmits to others the feelings he has lived through," and the audience is "infected" by them. Sincerity is the whole test. Skill, beauty, and fame are irrelevant — or worse, camouflage.

He applied the test without mercy. Shakespeare: bad. Wagner and late Beethoven: bad. His own novels, including War and Peace and Anna Karenina: bad — aristocratic art, made for the few, transmitting feelings the common person couldn't share. Good art, for Tolstoy, was either religious art uniting people in shared feeling or simple folk art anyone could receive. Almost nobody — including artists who admire the definition — has been willing to accept his verdicts. Which exposes the trap in every feeling-based definition: whose feelings, judged by whom?

Round Four: Duchamp Moves the Fight to a New Question

Marcel Duchamp's Fountain (1917) — a standard porcelain urinal, signed "R. Mutt," submitted to the Society of Independent Artists in New York — is usually called a prank. It was actually a legal argument. The Society's rule was that any artist who paid the fee could exhibit. The hanging committee hid the piece behind a partition anyway, and in doing so proved Duchamp's point: the definition of art was being enforced by institutions, not discovered in objects. Nothing about the urinal was skilled, beautiful, or feeling-laden. If context and intention could make this a candidate for art, then every earlier definition had been measuring the wrong thing.

Round Five: The Philosophers Give Up on the Object

Twentieth-century philosophy absorbed Duchamp's lesson in three steps. Morris Weitz argued in 1956 that the hunt for a definition was itself the mistake: art is an "open concept," and works resemble one another the way family members do — overlapping traits, no single feature shared by all. Arthur Danto's 1964 essay "The Artworld" argued that what makes something art is an atmosphere of theory and history that the eye alone can't see. George Dickie then formalized the point as the institutional theory: an artwork is an artifact that has been conferred the status of "candidate for appreciation" by someone acting on behalf of the art world.

Each step moved the definition further from the object — skill, beauty, feeling — and closer to the context: the gallery, the theory, the conferring institution. Critics objected that this was circular ("art is what the art world calls art; the art world is whoever deals in art"). Defenders answered that the circle was the honest description of how it actually works. Both were right, which is why the argument stalled rather than ended.

Round Six: The AI Rematch

Then the machines entered, and every old position was dusted off within a decade. In 2018, Christie's sold Edmond de Belamy, a portrait produced by a generative adversarial network, for $432,500 — against an estimate of $7,000–$10,000. In 2022, Jason Allen's AI-assisted Théâtre D'opéra Spatial won first place in the Colorado State Fair's digital arts competition, to furious objection from other entrants. In 2023, the U.S. Copyright Office issued guidance that purely AI-generated material, lacking human authorship, is not copyrightable.

Watch the old arguments reappear in new clothes. The skill objection is Plato's: no craft, no art. The feeling objection is Tolstoy's: a model has lived through nothing, so what is being transmitted? And the institutional answer is Dickie's: the work was entered, judged, exhibited, and sold — status conferred, case closed. The AI debate is not a new argument. It is the same 2,500-year argument with a faster opponent.

The Argument at a Glance

Why the Argument Never Ends — and Why That's Useful

Look at the pattern in the table: every definition is a perfect fit for the art of its own moment and a poor fit for the art that came next. Plato's imitation theory couldn't see abstract painting coming; Tolstoy's sincerity test can't rank two sincere works; the institutional theory describes the gate but not what makes anything worth letting through. Art keeps changing faster than the definitions built to contain it — and artists, not philosophers, do most of the changing. The definition is always a trailing indicator.

For a working artist, the stalemate is actually good news. It means no committee owns the word: not the academy, not the market, not the algorithm. What you can own is a position in the argument — knowing which tradition you're extending, and which one you're deliberately breaking, is what turns a body of work into a point of view. The definitions themselves, laid out for reference, are in Part 1: What Is Art — The Complete Guide.

Frequently Asked Questions

Why is there no single agreed definition of art?

Because art keeps changing faster than definitions can contain it. Each era's definition fits its own art and fails on the next era's — imitation theory can't handle abstraction, feeling-based tests can't rank sincere works, and institutional theory explains who decides without saying what should be decided. Philosophers like Morris Weitz concluded the concept is "open": a family resemblance rather than a fixed essence.

What did Duchamp's Fountain actually prove?

That the status of art is conferred by context and institutions, not found in the object's skill, beauty, or materials. The Society of Independent Artists had promised to exhibit any paid entry; hiding Fountain anyway showed the definition was being enforced by gatekeepers — which shifted the whole debate from "what is art made of?" to "who gets to say?"

What is the institutional theory of art?

Developed by George Dickie (building on Arthur Danto's "artworld"), it defines an artwork as an artifact that has had the status of "candidate for appreciation" conferred on it by someone acting for the art world — artists, curators, galleries, critics. Its strength is describing how art status actually gets granted; its weakness is circularity.

Is AI-generated work art?

It depends which round of the argument you side with. By the institutional test, AI work that is exhibited, judged, and collected already functions as art. By Tolstoy's feeling test or Plato's skill test, it fails. Legally, the U.S. Copyright Office's 2023 guidance holds that purely AI-generated material lacks the human authorship copyright requires — a separate question from whether it is art.

Did Tolstoy really condemn his own novels?

Yes. In What Is Art? (1898), Tolstoy judged War and Peace and Anna Karenina as bad art: aristocratic works transmitting feelings accessible only to a privileged class. His definition — art as sincerely transmitted feeling that unites its audience — left most of the canon, including his own masterpieces, on the wrong side of the line.