Why an MIT Math Professor Still Wants Students to Solve Problems Without AI

MIT professor John Urschel used AI to solve decades-old math problems but says students still need struggle to truly learn.

AI Tools
Why an MIT Math Professor Still Wants Students to Solve Problems Without AI

John Urschel, an assistant professor of mathematics at MIT, has used artificial intelligence tools to help solve math problems that had stumped researchers for decades. He described the experience in an interview with MIT News published on November 17, 2026.

Urschel recently posted two preprints on arXiv. One studies the growth factor of random matrices, and the other looks at worst-case behavior in Gaussian elimination.

He said the work combined human insight with AI tools. These were the first two papers in which he used AI in his research.

Before earning his PhD at MIT in 2021, Urschel played offensive line for the Baltimore Ravens. He now holds the title of Class of 1956 Career Development Assistant Professor in the Department of Mathematics.

Urschel said he has loved puzzles since childhood, including Sudoku, crosswords and Martin Gardner math puzzles. He said the process of working through a hard problem is what he enjoys most about math.

Working With a Machine Collaborator

Urschel said using AI felt like having a collaborator with seemingly unlimited resources. He said the system did certain tasks much faster than he could.

He compared the shift to how computers handle calculations faster than people. In his view, it is now clear that AI systems can do some types of math far more efficiently than humans.

Still, Urschel said he finds problem solving most joyful when he works alone, without AI. He said the true value of a machine-made proof comes from a human understanding it and explaining it to others.

He tells students who use AI to focus on understanding. For students who do not use AI, he focuses on building problem-solving skills.

Concerns About Learning Without Struggle

Urschel said AI can be a major win for education because it gives easy access to vast knowledge. He noted that neither of his parents were professors and that he learned advanced math in high school using a laptop and Google.

He also called the technology a double-edged sword. Students can now finish homework or write a 10-page essay with the push of a button.

According to Urschel, this can give students the false impression that they understand material when they do not. He said avoiding struggle can create the feeling of learning without real learning.

He said faculty also have work to do. That includes making recitations and office hours as useful as possible and reminding students to take responsibility for their education.

Urschel said learning how to think will be more important than ever. He said knowing facts may matter less when machines hold so much information, and the ability to think for yourself could be at a premium if that skill weakens over time.

He also pointed to community as a reason to study math in person. Learning with a group of peers, he said, improves the overall experience.

His comments follow a recent report from MIT’s Ad Hoc Committee on AI Use in Teaching, Learning and Research Training. That report stressed productive struggle and the collaborative side of learning.

Urschel said MIT is holding many conversations about these issues now. He said the way the school prepares students “absolutely cannot look like what it was 10 years ago.”

From our research desk
AI Jobs Automation Index
Which jobs are AI tools targeting most? We mapped 3,400+ AI tools to real job functions — with BLS employment & salary data.
Explore the index
maisiekooc

Written by

Maisie is a news writer at Agent Locker, covering the latest developments in artificial intelligence, emerging technology and the companies shaping the future.

Discover AI Agents