# A computer scientist-novelist reflects on AI and our understanding of maths > Source: > Published: 2026-09-21 06:17:26+00:00 *On September 8, OpenAI* *claimed* *it had solved the Navier-Stokes problem. “The question of whether smooth three-dimensional fluid motion can break down has remained unresolved for roughly 90 years,” Open AI said. It put 10,000 AI agents to work on it and the task took 88 hours. What might this mean for how we “do” mathematics?* It is not really clear how mathematicians chose to become mathematicians. After all, there are no “creative mathematics” courses for the budding mathematician the way there are “creative writing” courses for the budding storyteller or poet. There are no celebrity mathematicians to crush on (let’s ignore Brian Cox). Mathematics books for children have ominous reassurances of just how *fun*, just how *easy*, just how *natural* the subject is; this is in direct contradiction with the student’s own experience. So either the books are lying or the student is a duffer. Frankly, given the circumstances, it is a wonder anyone becomes a mathematician. Though we do not know how mathematicians are made, we do know what mathematicians *do* after they become professional mathematicians. They work on proofs. That’s really it. They prove things day in and day out, and *if* they have time left to spare or *if* their lunch money depends on it, *then* they’ll indoctrinate hapless graduate students to also prove things day in and day out. If this life seems charmless, that’s because it is. Until, of course, one proves a result so spectacular, so beautiful, so difficult, one can have it engraved on your butt. *That* moment makes it worth all the years of tears and salt. ### A long-standing question Such a result was achieved on September 8 when a long-standing question about the Navier-Stokes equation was finally settled by two teams. One team was composed of AI agents and the other by a couple of human mathematicians who made use of AI tech. They [showed](https://www.science.org/content/article/how-ai-math-breakthrough-ignited-controversy) – meaning, they *proved* – that the Navier-Stokes equation can exhibit singularities. The details of the achievement don’t particularly matter here. The Navier-Stokes equation describes the behaviour of a class of liquids. Since liquids misbehave in all conceivable ways, it is a gnarly beast. Claude-Louis Navier, who first formulated the equation in 1822, almost failed his École Polytechnique entrance exams. He also built a bridge that collapsed from the weight of all the assumptions he’d piled on the design. On the other hand, George Stokes, who provided the first reasonably rigorous derivation of the equation in 1845, was already getting cow-eyed by Oxbridge when he made Senior Wrangler in the notoriously difficult Mathematical Tripos exams. How did Claude-Louis Navier and George Stokes get interested in math? Claude-Louis Navier, I suspect, could just as easily have been diverted to write theses on literary criticism. Stokes’s interest in waves, in all the wonderful weirdness of liquids, seems to have originated in the idyllic childhood spent living very close to the Atlantic Ocean. His home was a mere 3.5 km from Dunmoran Strand beach in Skreen, Ireland. This makes for a nice story, but plenty of us live very close to garbage dumps and yet somehow fail to take an interest in the mathematics of garbage. All we can really say is that living close to the beach didn’t subtract from George Stokes’s mathematical inclinations. ### Proof of the matter What if we have our kids living, not near the ocean, but adjacent to a friendly AI? When they have a problem with fractions, the AI is there to help out. If it’s help with a doctoral thesis, why not? The AI is there to help with that too. At every step, the AI surrounds our young, nudging them, guiding them, infinitely patient, infinitely competent, and infinitely forgiving. Would kids still choose to become mathematicians? Or would they never get over their awe of their first and last teacher, the AI, and settle for viewing mathematics as a spectator sport? It’s tempting to argue that just as chess programs didn’t kill off human interest in chess, AI won’t kill off human interest in mathematics. The difference is this: any game of chess is independent of every other game of chess. So you can isolate games humans play with other humans from the games humans play with programs. Segregation brings with the illusion that some sort of parity has been achieved. That is not the case with mathematics. You cannot solve a problem in mathematics without it influencing other solutions. Humans cannot isolate the mathematics problems they solve from the problems an AI solves. For example, the fact that the two teams cracked the singularity problem means that going forward, mathematicians must accept that this result exists. If AI can “do” mathematics, then clearly, the way *we* do mathematics has to change. What does it mean to “do” mathematics? It is not, as OTT shows like *Prime Target* would have you believe, to be good-looking and have the ability to see π everywhere. Doing mathematics has come to indicate one very specific talent: the ability to prove things. Proofs have become the skeleton, bones, flesh and lifeblood of mathematics. Once a student leaves high school, they enter the world of mathematical proofs. Every claim they encounter in a math textbook has a proof. To know something in math is to have a proof for it. In fancy terms, knowledge is valid inference. It is this bedrock that will need to be rethought. AI will soon, if it is not already, become far better at developing proofs than humans. In fact, we will come to see that the miracle isn’t that machines can do proofs; it is that we are so good at it. What is left for humans to do? Plenty. The fact that we can outsource much of proof development to programs is one of the best things that could have happened to mathematics. Proofs have been getting impossibly unwieldy. As our mathematical problems get ever subtler, the landscape has gotten increasingly treacherous. We need a vehicle to help us fly over it and not have to trudge for years on end through thicket and turbulence. AI is that vehicle. For the past 200 years or so, mathematics has selected for a certain kind of mind: detail-oriented, obsessive, narrow-focused and immune to boredom. It is as if poetry were practised by only allowing poets with a passion for prosody. This move towards turning proof into a near-fetish object isn’t accidental. Nor was it unjustified. A great many mistakes were made by 17th- and 18th-century mathematicians because they didn’t understand the importance of rigour. It is just that in achieving rigour, through magisterial projects like those of David Hilbert or the Bourbaki school, mathematics became afraid of speculation and play. Mathematics books are devoid of humour, personality and any indication that a human mind contemplated the results. The texts are an endless succession of claims, proofs and exercises. The student is warned that to miss the exercises is to ensure failure as a mathematician. Definitions are unmotivated, history is almost always ignored, and alternate theoretical approaches, failed or otherwise, are rarely discussed. In the more abstract areas, any application to “real-world” problems is viewed with some dismay. It’s no use to appeal to one’s professors because they reveal only the survivor’s lack of compassion. As the broken scholar, now a tenured academic, approaches the twilight of their mediocrity, a certain frivolity is permitted: a note on the golden ratio in the local newspaper, some twaddle on the importance of intuition in creativity, or perhaps even – this is pushing their luck with the department chair – the need to reform the system. But there is another kind of mathematics and another kind of mathematician. This style is embodied by mathematicians like Leonhard Euler, Srinivasa Ramanujan and John Horton Conway, René Thom and the sparkling imaginativeness of their work. René Thom, a Fields Medallist, is especially interesting. Ivar Ekeland, in his obituary essay on Thom, wrote: “When I first met René Thom, in 1970, he was already a legend in mathematics. There were many stories about him, which were reverently passed around. It was said, for instance, that his thesis adviser, back in the early 1950s, was concerned that he could not get Thom to write satisfactory mathematical proofs; another eminent mathematician told the adviser not to worry– there are ten people in the world, he said, who can prove these theorems once they have been stated, but only Thom can state them. “Indeed, Thom's own lectures gave some credence to such tales. He was a geometrician; he saw things in his mind and would try to impart his vision with all the means at his disposal, whether they were pictures on the blackboard or analogies from other sciences. In French mathematics, he was unique in this respect. There was general agreement at the time that serious mathematicians were in the business of devising proofs, not explaining them. The meaning of a theorem should be worked out from its formal proof. With Thom, it was the other way around. He showed why things were true, or should be true, and left his audience to work out the proofs.” Perhaps this is the kind of mind which the new age of mathematics will favour. A mind which delights in speculation, leaps of the imagination, a deep sense of aesthetics, and a personal style which can even imprint how a field gets to be studied. It is mathematics approached as play. The philosopher Bernard Suits wrote that “playing a game is the voluntary attempt to overcome unnecessary obstacles.” It is precisely this voluntary acceptance of the lack of necessity which turns mathematics into art, and the kind of art which brings with it the joys of play. No doubt the day will surely come when the universe will see fit, just as it did with us, to turn to these engines of our devising to answer its own enigma, but for now, the responsibility of asking the right kind of wrong questions remains our prerogative. *Anil Menon is a computer scientist and a fiction writer. His* most recent work is the novel The Coincidence Plot. *It was preceded by a collection of his speculative short fiction,* The Inconceivable Idea Of The Sun: Stories. *He is currently the chief editor of* The Bombay Literary Magazine. *We welcome your comments at*