John Urschel PhD ’21, the Class of 1956 Career Development Assistant Professor in the Department of Mathematics, is no stranger to growing muscle: Prior to obtaining his PhD, Urschel enhanced his offensive capabilities as a lineman for the Baltimore Ravens. He continues to be dedicated to assisting his kids in physical activity, although his emphasis has shifted to critical thinking abilities. In the classroom, Urschel shares with his pupils his lifelong love for solving puzzles, while polishing their ability to be strong qualitative thinkers; qualities he says will be important in an artificial intelligence-enabled society.
Urschel recently published two preprints on arXiv, one on the growth factor of random matrices, the other on worst-case behavior in Gaussian elimination, in which he employed a combination of human understanding and AI technologies to tackle problems that had stymied mathematicians for decades. That encounter presented Urschel with new insights on the future of mathematics and higher education, and the enduring value of the in-person, residential education offered on college campuses, where students can experience the importance of struggle, problem solving, and human connection. Urschel recently chatted with MIT News about education, AI, and mathematics; extracts from the conversation follow.
What aspects of mathematics do you aspire to impart to your students?
A: In my early childhood, I had a passion for solving puzzles, including Sudoku, Martin Gardner’s mathematical conundrums, and crossword puzzles. I believe there’s something very enjoyable about being faced with some difficulty that you don’t fully comprehend, trying to find your bearings, trying to navigate your way through it, and finally gleaning some form of insight from it. That’s a process that I’ve appreciated for as long as I can remember, and that’s the thing that I’ve truly enjoyed most about math.
When it comes to education, initially, I want to help our students train themselves and develop their capacity to solve issues and their ability to be good quantitative thinkers.
The second half, which is more specific to whichever subject I’m teaching, is to help them integrate the key ideas or tools or strategies or themes that help them become better problem solvers in that specific discipline. If they’re taking a course on a specific subject, the aim is for me to try to transmit critical basics so that they can use that learned expertise to become better problem solvers out in the world.
Q: You recently employed AI to assist in solving two proofs. Could you elaborate on the experience?
A: These are the first two papers where I’ve ever handled the machine, and it does feel different. It’s not unlike having a collaborator, but it’s a collaborator who has access to endless resources, seemingly, and does some things much, much faster than I ever could.
In the same wa that computers can do computations far quicker than any of us can, I think at this time it’s fairly evident that artificial intelligence systems can do certain types of mathematics much, much more efficiently than any of us can.
Q: What was it like using AI to tackle arithmetic problems? Was it just as rewarding?
A: For me, personally, I would say that the sensation that I talked about of problem solving, of working your way through a challenge, I find that that’s most pleasurable for me when I do that by myself, without AI.
One thing that I believe is incredibly important is to have an independent sense of problem-solving ability, a sense of curiosity — this idea of building those mathematics muscles, which I think are necessary, even in the era of AI.
Particularly when conversing with a machine and requesting information, I believe it is crucial to enhance the concept of comprehension. If you’re talking to a machine, let’s say, and it gives you a proof of some mathematical fact, I think the true value in that proof comes from a human understanding and assimilating the way in which the machine combines certain mathematical ideas, and then internalizing that and communicating that knowledge to others.
Understanding is the key thing that I highlight to my students who use AI systems. And then for the kids who do not use AI systems, I focus on strengthening their problem-solving ability, which is what we’ve done for centuries in mathematics.
I think in the age of current AI machines, as long as there is a role for humans to think and try to solve problems, which I believe there is, then having these abilities remains crucial.
Q: The recent study from MIT’s Ad Hoc Committee on AI Use in Teaching, Learning and Research Training underlined the necessity of experiencing the essential and productive struggle of learning, coupled with the collaborative side of learning. I’m curious what you think about the role of AI in math instruction going forward.
A: On one side, the advancement of AI technologies needs to be regarded as a significant triumph for education, a substantial victory for equalizing opportunities when technology is readily available, providing access to extensive knowledge.
Neither of my parents is a professor; I remember when I was a kid, I didn’t know what a mathematician was until I got to college. And the way I learnt about sophisticated mathematical topics in high school, and when I came to college, was on a laptop, and with Google. I was able to Google some issue I wanted to learn about, and it was fantastic that I could find papers published by some of the world’s finest experts, just freely available online at my fingertips to learn from. That was exceedingly significant for me and immensely beneficial for my learning.
However, I believe we have reached a juncture when contemporary technology presents a duality of benefits and drawbacks. On one hand, AI systems enable accelerated learning, rendering the resources available to me in my youth trivial in comparison to the contemporary tools that facilitate self-education.
However, the concern is that this may also diminish your capacity for independent thought and learning. I believe it is challenging for both high school and college students. You are overwhelmed and fatigued, and with the assistance of AI, you may effortlessly do your assignments at the touch of a button. You can compose your 10-page dissertation with an only click. Moreover, one may get a misleading sense of comprehension regarding the situation when, in fact, one does not. I believe that is one of the significant risks when one fails to pursue knowledge independently.
People don’t like the word “struggle.” But when you don’t struggle with a given thing on your own, I think you can obtain the sense of learning without actually learning. Engaging with the topic and challenging oneself is a significant ability.
Undoubtedly, there is a necessity for faculty and instructors to enhance the utility of recitations; to ensure that office hours effectively assist students; to motivate students and remind them of their accountability in their education, emphasizing that active participation and engagement with the material is paramount. We must acknowledge our obligation to provide students with the optimal learning experience and to offer an educational framework that both challenges them and is a high-quality product they are eager to engage with.
Q: What rationale would you provide for motivating students to continue their education in mathematics despite the progress achieved by AI technologies? Do you still see the future of mathematics as exciting?
A: Indeed, for numerous reasons. The process of acquiring critical thinking skills will become increasingly vital. The notion of being an authority, with extensive knowledge that is readily accessible, may be somewhat diminished in the presence of a machine equipped with abundant resources. I believe that the capacity for independent thought may grow increasingly significant. The presence of these devices engenders a legitimate concern that this skill set may diminish among the general populace as generations progress, making that capacity increasingly valuable. I believe that education in mathematics and quantitative disciplines is essential for it.
The subsequent aspect pertains to community. I believe that acquiring knowledge in a collaborative environment, rather than in isolation, significantly enhances the whole experience. It constitutes a fundamental aspect of the experience. I believe those two factors underscore the significance of studying mathematics and quantitative sciences, as well as the necessity of attending class.
The globe is undergoing significant transformations at a swift pace. I believe it is our duty as an institution to equip students for it. We are currently engaged in numerous discussions around this matter, as the capacity to think, reason, and resolve issues is a fundamental competency that our kids will increasingly require. I believe it is essential to ensure that we adequately prepare our pupils for the world they are about to join. The methodology we employ to educate our pupils, or equip them for the future, must fundamentally differ from that of a decade ago. This planet appears markedly dissimilar.

