AI Is Not Going to Kill My Love of Math
21 days ago
- Recent AI-assisted proofs, including a Navier-Stokes counterexample, raise deep existential and philosophical questions about the future of mathematics.
- The author, a hobbyist mathematician, feels no excitement at the breakthrough, but rather anxiety about AI's long-term impact on research.
- Short-term wins from AI proofs are real (e.g., counterexamples), but the long-term consequences for mathematical research remain unclear.
- Terence Tao warns that premature AI proofs may 'contaminate' problems, skipping the fruitful intermediate steps that generate new sub-problems and insights.
- The author partly agrees but notes that applied math may benefit from fast results, and the universe likely contains unbounded mathematical depth, so problems won't run out.
- AI-generated proofs often lack beauty, conciseness, and human interpretability, though future scaling might change that.
- There is a risk that young people lose motivation to study math if AI solves everything, but the author remains personally committed to the joy of learning.
- Even if AI dominates research, humans may still find value in doing math for leisure, like competitive chess after computers surpassed humans.