This morning, my op-ed about artificial intelligence, math, and chess was published in the Washington Post. Here’s a link, but (warning!) the article may be behind a paywall. Here’s my brief synopsis.
Mathematicians have been facing an unprecedented summer of reckoning, as artificial intelligence has proved itself capable of solving major unsolved problems. Over the last three years, we have clung to an increasingly tenuous belief that there are certain kinds of problems (ones that require “creative thinking”) that humans can still do better than AI’s. The solution of the Navier-Stokes problem, in my opinion, removes the last illusion of human superiority. We have now entered the age when AI can now solve any math problem as well as humans. (There are some caveats to that statement. I would say that they are still not as good at explaining their solutions; they are roughly equal to the best of us in finding solutions; and they are WAY WAY faster than us.)
My Washington Post op-ed looks at these developments from the point of view of a chess player. In the world of chess, we already had our moment of reckoning long ago, in 1997, when Deep Blue beat Garry Kasparov in a six-game match. But surprisingly, that wasn’t the end of chess. In fact, it has kicked off a golden era. You might think that grandmasters would have become irrelevant. Think again! In the age of streaming video, they have become media stars. Chess has never been so popular, and I argue in the Washington Post that computers with superhuman abilities have been a large part of that boom. They make top-level chess more accessible. They democratize it. I see this as a possible path for mathematics, assuming that mathematicians can adjust to the new landscape, and not fall into the “woe-is-me” trap.
AI has been on my mind a lot lately. Two weeks ago, I wrote an article for Science News about another math problem (the small prime gaps problem) in which AI had set a new “world-record.” The previous record had been set by a human three days before, who rushed to post her solution on the Internet before OpenAI could post their improvement (which was then rumored but hadn’t been announced yet). Julia Stadlmann will almost certainly be the last human to ever hold this record.
The prime gaps story actually broke five days before the Navier-Stokes story, but the latter story was much bigger news. I fear that Julia’s accomplishment got lost in the shuffle to some extent, but I appreciate the fact that Science News still saw some value in my article.
AI also puts a new spin on my book, Master Sun’s Problem, which will be published next month. At first, I thought that my book might just end up being a requiem for math as it used to exist, pre-2026. But upon further reflection, I realized that my book speaks to an important divide in the math community. For some people, and for some mathematicians, math is about solutions. That’s what you learn in math class, right? The teacher asks you questions, and you’re supposed to find the solution.
But I emphasize a different point of view in my book: for me — and for many mathematicians — mathematics is about the journey. It is what you learn on the way to the solution that matters the most.
If you are in the camp that believes the solution is the only thing that matters, then I can certainly see why you would be worried by the latest AI developments. If the computer can find answers more efficiently than humans, then why would we need human mathematicians any more? On the other hand, if you believe that the value of math lies in the struggle and in the deeply personal experience of coming to understand something that was previously a mystery, then mathematics will continue to be a source of joy and delight, and mathematicians will continue to have a purpose.
