Noah Smith
· Noahpinion
· 2026년 8월 4일, 07:36
· ⏱ 18 분 소요
| Substack에서 읽기 ↗
요약
Frontier AI has reached the point where it can independently solve major open problems in mathematics, ending the era in which individual human geniuses were the only possible discoverers of new theory. For markets, the article is another validation of frontier model capability—especially OpenAI's Astra—though it contains no explicit buy or sell recommendations.
•OpenAI announced that its new model, Astra, managed to solve ten major outstanding problems in math and theoretical computer science, with consensus calling them extremely important, stunning breakthroughs.
•University of Toronto mathematician Daniel Litt, a prominent AI-math skeptic, conceded that AI's inability to do certain kinds of high-quality research is likely a matter of 'not too much' time.
•Mathematician Kirwin Hampshire's viral post described a 'Dark Night of Mathematics,' citing LLM counterexamples to significant long-standing conjectures and a profound spiritual crisis among researchers.
•Fields Medal winner Jacob Tsimerman took a job at OpenAI and said he does math 'largely because I enjoy it and I am good at it,' warning that if the challenge of problem-solving disappears, many mathematicians might not want to continue.
•Noah Smith estimates U.S. academic mathematician salaries total roughly $4-5 billion—'a rounding error' compared with GDP, similar to spending on bowling alleys or Halloween candy—so society can afford to keep mathematicians employed as interpreters of AI discoveries.
•The article frames humans as 'edge compute' for a 'world-mind': society previously integrated individual human findings, while AI is now becoming the coordinating world-mind that uses all human knowledge as training data.
•AI is expected to help most scientists by removing math as a bottleneck: economists and other researchers can focus on data and empirical results instead of writing theory sections, and Przemek Chojecki is quoted as saying AI could boost science progress 100x.
•Historical heroes like Newton, Perelman, and Wiles were necessary because ideas had to be comprehensible and communicable to a single human brain; AI removes that single-human bottleneck.