Mothers, Do Let Your Babies Grow Up to Be Mathematicians

Last week I was mulling writing about the awarding of this year’s Field Medals, which are often described as math’s equivalent of the Nobel Prize (no doubt to the dismay of both awarding organizations). Nah, I told myself; sure, I’m interested in math, but how many other people are? Most people barely even care about Nobel Prizes, and when they do it’s likely to be more about the Peace or Literature Prizes, not Physics or Chemistry. Math reminds most people of junior high algebra.

Don't let the tuxedo fool you. Jacob Tsimerman winning the Fields medal. Credit: Simmons Foundation

But then I saw the hook: one of the medalists – Jacob Tsimerman, of the University of Toronto – announced he was "pivoting toward AI safety” and would soon be going to work at OpenAI. People can’t get enough about AI, and with AI agents “going rogue” and creating all sorts of mischief, AI safety should be top of mind for everyone interested ion our (AI) future.

Professor Tsimerman won the award for “his contribution in the recasting of o-minimality as a fundamental method of arithmetic and complex algebraic geometry, and his role in the proof of many central conjectures including Griffiths' conjecture on the algebraicity of images of the period maps, and the Andre-Oort conjecture for Siegel modular varieties.” If you have the slightest idea what any of that means, you may be going to work for an AI company soon too. Note, though, nary a mention of AI in all that.

He did, though, co-author a paper last year about various scenarios where AI lead to most or all humans being killed, so o-minimality may have to wait. That paper ominously concluded: An AI-driven omnicide may involve an intent to kill a large number of people, or may be wholly unintentional. And, if an intent to kill exists, that intent may exist within a large group of individuals such as a military or corporation, or just a few people, or in an AI system itself.”

He told Janet Hurley of The Toronto Star: “The reason I wrote that paper with Andrew Critch is that, even though this has been going on for a long time, I felt there wasn’t a visceral story people could engage with in terms of, concretely, what are we afraid of here?”

Consider me afraid.

It turns out that Professor Tsimerman is not alone in leaving mathematical academia for AI. There has been what has been described as a “wave of top mathematicians "exiting" academia for industry.” If you are a graduate student in math, or, better yet, a math professor, you are, as Ben Affleck’s character told Matt Damon’s math genius character in Good Will Hunting, “sitting on a winning lottery ticket.” The AI companies want you, and are putting up big bucks to get you.

Just in the past week Julia Amann of The Wall Street Journal writes of The Million-Dollar Talent Wars for 20-Something Math Geniuses and Lila Shroff of The Atlantic says Something Weird Is Happening in Math, both describing the impact of AI on academic mathematics.

Wall Street firms having been going after math students since the 1980’s to serve as “quants,” designing ever more elaborate trading models, while Silicon Valley firms have been snapping up computer science students for the last three decades. Now the AI companies can’t get enough math geeks either, and they have money to burn.

Matt Stabile, founder of New York-based recruitment firm Stabile Search, told Ms. Amann: “But a million dollars is something people don’t even bat an eye at anymore…The delineation is pre-OpenAI and post-OpenAI, that’s when you saw competition really take off.”  Charlie Witmer, COO of quant firm Optiver, acknowledged to her the effect of AI firm’s interest in math geniuses: “Outstanding people are in more demand than they’ve ever been, and that does create competitive pressure on wages. As competition increases for outstanding people, prices naturally rise.”

Part of the reason mathematicians may be tempted to leave academia is that, well, AI may soon be better than they are at math. A few months ago, I wrote about how mathematics might be one area that AI hadn’t yet conquered. That didn’t age well. “In a few years, AI systems will be robustly superhuman at the act of doing mathematics,” Professor Tsimerman told Ben Cohen of The Wall Street Journal. “The social consequence of that, how we choose to react, what you feel about it—those are much harder questions.”

He admitted to Ms. Shroiff that there are pluses and minuses (pun intended) to AI’s improving math skills:

From one point of view, I think it will be extremely exciting. We might speed up the process of generating interesting mathematics by enormous factors of 10 or 100. If that happens, we might see the connection between pure math and applications (which typically takes many decades) really speed up and become a much tighter pipeline.

But from the perspective of research mathematicians, and especially young people who are pursuing a Ph.D. in mathematics, it’s a bit of a turbulent time. The skills that we’ve acquired and learned to propagate might become less relevant than they are now.

On the other hand, he thinks math expertise may help us understand what AI is up to:

Mathematics is historically the language by which you take intuitions and fuzzy notions of how things work and you make them precise. We’ve done this with information theory; we have done this with complexity theory. Just a little bit of understanding and precise definitions can provide a ton of mileage. Once we have that understanding, the hope is that we could do a better job of anticipating the behavior of new AI systems, adjusting for them, controlling them, and reacting to them. That’s where I think mathematicians fit in.

He told Kenneth Chang of The New York Times about AI: “It’s mostly an empirical and engineering kind of science. If we understood how these things work better, we might be able to steer them better, or understand them better, or control them better.”

We can hope so, anyway.

The AI industry first sucked up university AI researchers, then academic leaders in computer science, and now is going after the math departments. The trouble is, there aren’t enough of them. Even worse, we’re eating our seed corn. Who will train the next generation of mathematicians?  That’s fine if you assume AI is that next generation, but if you still want human creativity and brilliance, it’s daunting.