If AI proved the Riemann Hypothesis, would I care?

Or: AI-assisted mathematics is the end of my interest in mathematics...

Most people these days know me as a wildlife photographer, but before my photography career, I was a research mathematician for seven years. I don't talk much about math these days, but recent discussions in the mathematical community prompted me to write this essay.

The discussions are about the role of AI in mathematics. A lot of mathematicians are saying that AI might assist with mathematical discovery. Some are even saying that if some AI model comes along that can prove theorems, there'd still be work to do and fun to be had because after all, the purpose of mathematics is not just to obtain results, but understand them.

Actually, this sentiment has become quite common among mathematicians, or at least among the vocal ones. It's also a common sentiment among many technophiles who otherwise have virtually no understanding of mathematics beyond rudimentary linear algebra. And, it's a sentiment that I don't share.

To be concrete, if AI happened to prove the Riemann hypothesis, I wouldn't care. And I find the Riemann hypothesis quite interesting. In fact, I was fascinated by this problem for many years ever since I heard of it. But AI finding a proof? Not interested. Ditto with any of my favorite conjectures about which I'll spare you from reading.

Now, it's absolutely true that mathematics is about understanding results – new understanding is the core of mathematics. And from that perspective, if AI came up with a clear and well-written proof of the Riemann hypothesis, it would increase my understanding by reading it. But mathematics is a lot more than just understanding.

Mathematics is also about the human journey. It's about having pride in your results derived after hours of sitting with a pencil and paper. It's also about the shared experience of sitting with others who enjoy the journey as well. An AI who can traverse the circuitous routes of logical deduction and vomit out elegant explanations bypasses that human element and takes away the magic.

And an AI proof of the Riemann Hypothesis would be exactly that: a proof without magic. It would lose its appeal to me because the community that collectively created it would no longer be a group of those passionately interested in math. Instead, the community that created it would be a factory created from the subtle shift in emphasis from the journey of mathematics to soulless industrial production.

In some ways, AI is not the source of the problem. Mathematics for some time has already been shifting towards industrial production due to its reliance on the modern academic system, which has itself been slowly pushed into complete subservience to the technological system. Even before AI, curiosity already was long ago demoted below the goal of sustaining large research programs of endless specialization that optimize obtaining research grants. AI is just the apex of that process, serving two functions: not only does it make the industrialization of mathematics much more efficient, it also promises to revitalize mathematics from the very staleness brought on by the industrialization process itself. Of course, AI tools will not be devoid of utility for the modern mathematician. They will prolong the usefulness of the human being in mathematics for a short time, until AI can take over completely.

And, I suppose that this industrialization problem is not unique to mathematics, either. In photography, we optimize for immediate visual impact and quantity – the sharing of experience no longer matters. In journalism, we optimize for clicks and ad views, instead of telling stories and making the world a better place. The point is that every human endeavor is mutating into some sort of industrial process that optimizes for just a few variables, usually negating human experience, and so of course AI is the natural next step. But do we ever ask whether that process is a good thing in the first place? No. We are too busy optimizing.

It's a sad thing. The human element is being stripped from human striving, and AI is the technical realization of this process, representing the apex of our evolved obsolescence. So that AI-generated proof of the Riemann hypothesis? I'd be happy to watch it burn to ashes without reading a single word, because the very process that led to its creation is what renders its content completely detached from what I value.