"Egoless Mathematics": A Guest Post on Terence Tao's Blog Asks What Maths Is For in the AI Era
Three authors argue on Terence Tao's blog that AI should push mathematicians to value understanding over output, rank or credit.
A guest post on Terence Tao’s blog, written by Karthik C. S., Amit Sahai and Jie Xue, climbed Hacker News on Monday. Its question is not “can AI do maths?” but “what is mathematics for, if AI can do much of it?” Their answer is an idea they call egoless mathematics: pursue truth without letting the discovery serve your own importance.
The authors argue that the problem is older than AI. Mathematicians, like many professionals, have come to measure their worth by output and individual ability: papers, citations, prizes. Those numbers are useful as signals but harmful when they define what the work is worth. Retreating to “AI can’t be creative or intuitive” is also shaky, they say, if creativity is valued mainly as an advantage over machines.
Instead they point to understanding. A formally correct result is not automatically illuminating, and good mathematics reveals why something is true or what connects two ideas. The joy of understanding, they write, does not shrink if someone else got there first. Two lines stand out: “This is a good time for mathematicians to think carefully about the purpose of their work,” and “Mathematics was beautiful before any of us were born, and it will remain beautiful long after we are gone.” They admit they are “still far from having attained such egolessness,” presenting it as an ideal, and they do not try to settle policy questions.
What’s behind it: This is an essay, not a study, so there are no numbers to check. It matters because AI tools are now used for proofs and research ideas, and professions are asking what remains distinctly human. The authors suggest that people who embrace AI and people who abstain can still agree on the goal of truth. That stance is thoughtful, though other mathematicians may well disagree, especially about credit, jobs and the role of verification.
What this means for you: You do not need to be a mathematician to use this. If AI helps you write, code or study, ask yourself whether you are seeking the result or the understanding. For learners, a good rule is to use AI to explain steps you do not follow, then try the next problem alone. For everyone, it is a calm contrast to the usual “humans versus machines” framing.
Sources
Source: https://terrytao.wordpress.com/2026/11/02/egoless-mathematics-in-the-ai-era/
Adobe Tests an AI Agent That Edits Images Step by Step, and Checks Its Own Work
Adobe Research and UC San Diego present EditAgent, which splits edits into steps, tries several versions and picks the best one using measurements.
