09-20-2026, 03:23 PM
In “Why Do We Need Human Mathematicians Anymore?”, Po-Shen Loh argues that the rise of increasingly capable AI forces mathematics—and every other profession—to confront a deeper question: what role should humans retain when machines can increasingly perform the technical work? He proposes a guiding principle: humans should use technology to promote human flourishing.
His argument is that as AI becomes more capable and autonomous, society will actually need more highly skilled humans capable of supervising, steering and making decisions about increasingly complex AI-driven systems. In mathematics, this means maintaining communities of human experts who understand the frontier deeply enough to decide which problems are worth pursuing, interpret results and direct AI toward socially valuable goals.
For mathematics specifically, Loh argues that AI should not be viewed simply as a replacement for mathematicians. AI may eventually generate enormous quantities of formally verified mathematics, discover useful theories and even surpass humans at proving individual theorems, but humans would still be needed to choose research directions, develop mathematical taste, understand significance and connect mathematics with human needs.
He points to the historical example of linear algebra, whose abstract development eventually became fundamental to computing, machine learning, GPUs and quantum mechanics. His conclusion is therefore not that mathematicians should reject AI, but that they should embrace it while preserving a pipeline of human mathematical expertise capable of steering it—and reconsidering how teaching, research, practical impact and human understanding are valued within mathematics.
ARTICLE
His argument is that as AI becomes more capable and autonomous, society will actually need more highly skilled humans capable of supervising, steering and making decisions about increasingly complex AI-driven systems. In mathematics, this means maintaining communities of human experts who understand the frontier deeply enough to decide which problems are worth pursuing, interpret results and direct AI toward socially valuable goals.
For mathematics specifically, Loh argues that AI should not be viewed simply as a replacement for mathematicians. AI may eventually generate enormous quantities of formally verified mathematics, discover useful theories and even surpass humans at proving individual theorems, but humans would still be needed to choose research directions, develop mathematical taste, understand significance and connect mathematics with human needs.
He points to the historical example of linear algebra, whose abstract development eventually became fundamental to computing, machine learning, GPUs and quantum mechanics. His conclusion is therefore not that mathematicians should reject AI, but that they should embrace it while preserving a pipeline of human mathematical expertise capable of steering it—and reconsidering how teaching, research, practical impact and human understanding are valued within mathematics.
ARTICLE
┌────────────────────────────────┐
│ KONSTANTINOS MICHAILIDIS │
└────────────────────────────────┘
│ KONSTANTINOS MICHAILIDIS │
└────────────────────────────────┘