The Enduring Value of Math, in an Age of AI

[mklabgr]

The Enduring Value of Math, in an Age of AI
Author: Francis Su
Published: August 20, 2026
Topic: Mathematics education, AI, mathematical thinking

Francis Su addresses a question that is becoming increasingly urgent as AI systems grow capable of constructing proofs, finding counterexamples, and making genuine progress on difficult mathematical problems: why should humans continue studying mathematics if machines can eventually perform many mathematical tasks better than we can? His answer is that mathematics education has never been merely about acquiring technical skills such as solving equations, applying theorems, or writing proofs. Its deeper purpose is to cultivate what he calls mathematical virtues—creativity, persistence, intellectual flexibility, curiosity, the ability to recognize hidden structure, comfort with uncertainty, and the willingness to struggle with difficult problems. Skills can become obsolete or automated, but these habits of mind become part of the person doing mathematics. 

Su therefore distinguishes between skills—things we know how to do—and virtues—ways of thinking and being. AI may increasingly perform the first category, but humans will still need the second to determine which questions matter, recognize whether an AI-generated result is significant, connect discoveries to broader ideas, and decide how mathematical technologies should be used responsibly. In mathematical research, he expects AI to become a powerful research assistant. This could even broaden participation: someone with strong intuition or an excellent research question but weaker technical ability might use AI to develop ideas that previously would have been beyond their reach. 

The implication for education is that assessment should change as well. Su plans to reward good strategies, persistence, insightful questions, collaboration, and genuine understanding, rather than judging students solely by whether they reach the correct proof. His central argument is ultimately humanistic: an AI can produce an answer or even a beautiful proof, but it cannot give a student their own understanding of that mathematics. The struggle of being stuck, experimenting, discovering a connection, and finally understanding why something is true is itself part of mathematics' value. Thus AI does not make learning mathematics pointless; paradoxically, it may reveal more clearly why mathematics was worth learning in the first place. 

Key takeaways

• Mathematics education is not primarily about producing answers; it develops ways of thinking.
  
• As AI automates mathematical skills, human qualities such as creativity, judgment, curiosity, persistence, and the ability to ask important questions become more valuable.
  
• AI could transform mathematicians from primarily proof producers into question askers, interpreters, evaluators, and directors of mathematical exploration.
  
• Su's most important distinction is perhaps: AI can give you an answer, but it cannot give you your own understanding. 
  
  

My assessment: This is a particularly important argument for mathematics teachers. If AI becomes extremely good at solving textbook problems, then an educational system built mainly around “solve these 20 exercises correctly” becomes increasingly difficult to justify. The educational value shifts toward why a method works, how a student approaches an unfamiliar problem, what questions they ask, whether they can evaluate an argument, and what mathematical insight they gain from the process. In that sense, AI may force mathematics education to move closer to what mathematics education ideally should have been all along.

ARTICLE