How AI does, and does not, change the way I do math

Rachel Webb says LLMs will alter the mechanics of her math research without changing the standard of mathematical interest. She still wants understanding, beauty, and interestingness, especially in relation to applications, open problems, and other math.

She gives two human reasons to keep doing math: it is enjoyable in itself, and it builds community through shared work and conversation. AI may speed up research and affect collaboration, but she argues it should be used in ways that preserve understanding and human connection.

Reading notes

#

• Research is a search for something interesting to say, and in math that means finding ideas that are clear, correct, and interesting.
• Mathematical work has a creative side: definitions, lemmas, and theorems can be arranged strategically or artistically to capture truth.
• AI may change execution by giving access to standard lemmas and speeding up research.
• LLMs do not change her criterion for interest, which still includes applications and proximity to major open problems.
• One human reason to do math is that it is personally enjoyable, and AI should be used only when it increases understanding as well as convenience.
• Another human reason to do math is that it creates communities through students, collaborators, conferences, colloquia, and informal talks.
• AI could weaken these bonds by increasing scooping fears or making it easier to ask a machine than a colleague.
• The post ends by calling for an age of AI that also keeps attention on the mathematical neighbor.