Ask two colleges what degree a mathematics major earns. A liberal arts college stamps Bachelor of Arts on the diploma. A technical school stamps Bachelor of Science on the same coursework. Pick a third college, and the letter flips again. The letters differ. The mathematics does not. A registrar chose one letter, and a different registrar chose the other. The choice tracks a school’s self-image, not the subject’s content. No theorem voted.

We think mathematics is an art. We hold this as an opinion, and we will tell you why we hold it. Verifiable facts come first. Our interpretation wears its label. The diploma does not decide the question, so we will weigh the evidence ourselves.

Two thousand years on the arts shelf

For roughly two thousand years, mathematics sat inside the liberal arts. The quadrivium held arithmetic, geometry, music, and astronomy. The trivium held grammar, logic, and rhetoric. Together the seven studies aimed to free the mind. A free person studied them because freedom needs training. Students pursued the arts for liberation, not for use. The old teachers did not ask what geometry did for a harvest.

People filed mathematics with the sciences in the nineteenth century. The twentieth century kept the habit and built departments around it. A clerk made that choice, not a philosopher. Two thousand years of shelf order say otherwise.

The experiment and the proof

Here is our reasoning. Science answers to experiment. Observation revises its claims, and nature’s facts stay contingent. Mathematics answers to proof. A proof yields a necessary truth. No observation can overturn one. No experiment can unseat the prime number theorem. Nature permits a different physics. Nothing permits a different arithmetic. Science writes its theories in mathematics. A language is not one of its own sentences.

The honest paragraph

Reasonable people disagree with us here. The Quine-Putnam indispensability argument makes the strongest case against us. Quine and Putnam pressed it in the twentieth century. It holds that our best theories need mathematics, so evidence for them supports the mathematics inside them. Physics rests its mathematics on the same evidence, the argument goes. That is a serious position with serious defenders. We do not dismiss it.

Our answer stays short. Indispensability shows that science uses mathematics. It does not show that mathematics is a science. A ledger serves accounting, and a ledger is not an accountant.

How mathematicians grade

Watch how working mathematicians judge their craft. They praise elegance. They praise depth. They praise beauty. No laboratory can falsify a theorem for inelegance. Nobody demands a control group for a proof. A physicist fears the anomaly. A mathematician fears the ugly proof.

G. H. Hardy wrote a famous apology for his life in number theory. He was proud of its uselessness. In it he said, “I have never done anything ‘useful’.” He considered that a credential, not a confession. Those are criteria of art, not of falsifiability.

What fell, and in what order

Now look at the machines, and look at the order of their conquests. They learned the arts first. They wrote passable prose. They drew convincing pictures. They made songs, then video. The public argued about taste. Nobody ran a trial. Taste referees those domains, and the world stays out of the judging.

Then mathematics went with the arts. In 2024, Google DeepMind’s systems reached silver-medal standard at the International Mathematical Olympiad. One of them, AlphaProof, wrote proofs that formal proof software checked, line by line. In 2025, Gemini Deep Think, from the same lab, reached gold-medal standard at the same olympiad. In the same month, a startup called Harmonic announced gold-medal-level results with proofs a machine verified in full. Silver, then gold, inside two years.

Science still stands. The same systems draft code and propose hypotheses. A human still watches them work. Science’s referee is the world, and the world does not grade drafts. Read this as observation, not as proof. The machines did not settle our taxonomy. They revealed a clustering. We find the order more interesting than any single result. Mathematics fell with the arts. It did not fall with the sciences. That is a datum about where the subject sits. It happens to match the shelf the quadrivium chose two thousand years ago.

The complication we owe you

One honest complication remains. Mathematics is the only art with a perfect mechanical referee. A proof checker needs no human taste. It certifies a proof on its own. So machine mathematics runs unattended, while machine code in a laboratory needs a watcher. But a referee governs the checking, not the making. Chess has a perfect referee. Chess did not become a science when the machines arrived. The referee settles legality, never worth. In mathematics, the choice of what is worth proving stayed human. That choice is the art.

What this means for a student

Suppose you weigh a mathematics degree. You should know what you choose. You choose an art with an unusually strict meter. The machines took the computation. They left you the choice of question. Students ask us, “When will I use this?” Our favorite answer: the way you use Shakespeare. You use it to become more free. You do not learn mathematics to out-compute machines. You do not learn writing to out-print a printer.

The empress dowager

We said this as an opinion at the start. We repeat it as an opinion at the end. We think mathematics is an art. An early biographer recorded Gauss calling mathematics the queen of the sciences, and number theory the queen of mathematics. Our view holds that the first title missed. Mathematics is not the queen of the sciences. It is the empress dowager of the arts.

If the bigger question finds you, we built tools for it. Start with the questionnaire, or face the mock defense.