All articles tagged with #mathematics
OpenAI says its Astra model solved 10 long-standing mathematics problems, stirring excitement about faster discovery and new techniques—while experts warn about credit attribution, access, funding, and how AI could reshape the future of mathematical research.
Generative AI is reshaping mathematics by accelerating discoveries across geometry, cryptography, and coding theory (OpenAI’s Astra highlighted with ten advances). This has sparked debates on attribution, correctness, and what constitutes mathematical creativity, as AI-assisted work floods journals and arXiv. Mathematicians diverge on how to integrate AI: some ethically exclude it, others embrace it as a collaborator. Case studies show both paths—one team avoided AI, another used it to achieve in 43 hours what previously took years—leading to joint papers. The field is grappling with how to reconcile AI's capabilities with transparency, integrity, and human collaborative values, signaling a nuanced future for math where machines help but human discernment remains central.
OpenAI previews Astra, its next major AI model designed to tackle long-running, multi-step problems; internal work claims ten significant advances in mathematics and theoretical computer science, with proofs formalized as Lean certificates, and potential release as GPT-5.7 or GPT-6 under a tiered access policy.
John Pardon, renowned for turning hard problems into unifying frameworks, has earned the 2026 Fields Medal for ground‑breaking work in knot theory, topology, and symplectic geometry—ranging from showing arbitrarily large knot distortion to resolving the MNOP conjecture on Calabi–Yau manifolds—and for developing new tools like virtual cycles and what’s now called the Pardon algebra, which are reshaping the foundations of symplectic geometry.
Two Chinese mathematicians, Hong Wang and Yu Deng, won the Fields Medal—the math world’s top prize—alongside John Pardon (US) and Jacob Tsimerman (Canada), marking China’s first Fields Medalists and Wang as only the third woman to win; each laureate receives CAD 15,000. Wang helped solve the 1917 Kakeya pencil-rotation problem in 3D, while Deng addressed Hilbert’s Sixth Problem by linking microscopic particle behavior to macroscopic gas dynamics. Wang studied at Peking University, MIT, and in France, and now teaches at NYU and IHES, while Deng is a professor at the University of Chicago; the prize is awarded every four years to under-40 mathematicians. Macron congratulated Wang.
Hong Wang and collaborator Joshua Zahl proved the 3D Kakeya set conjecture, a landmark result at the crossroads of harmonic analysis, geometric measure theory, and PDE, earning Wang the 2026 Fields Medal—the third woman to win the prize. The award caps a remarkable journey from Guilin to France, MIT, and NYU, and showcases Wang’s use of multiscale analysis and innovative “sticky” Kakeya techniques to rule out counterexamples, reshaping the field and pointing to future work on higher-dimensional Kakeya problems.
A mathematician, Levent Alpöge, used Anthropic’s Fable 5 to claim a counterexample to the Jacobian conjecture, a nearly 90-year open problem in algebraic geometry. Experts warn that a single counterexample, while notable, is not equivalent to a full proof and may offer limited insight, though the episode underscores AI’s potential in assisting mathematical research and discovery, with discussions also touching on OpenAI’s related progress.
A wall inscription at the Maya site of Xultun in Guatemala names the 8th‑century mathematician‑astronomer Sak Tahn Waax and encodes a 2,920‑day calendar cycle that links five Venus cycles with eight solar years and related Maya time units. Reported in Antiquity, Text 19 shows a complex, previously unseen calendrical math and suggests the chamber was a scribal workspace where mathematicians could be credited alongside artists in Maya society.
Researchers analyzing microtexts from the Xultun site in Guatemala have credited a specific Maya scholar, Sak Tahn Waax (“White-chested Fox”), with a unique mathematical formula that maps Venus and other planetary cycles to the 260-day ritual calendar and the solar year. This is the first time a Classic Maya math/astronomy work has been attributed to an individual, shedding new light on Maya science and its place alongside contemporaries in India, Iraq, China, and Greece.
Archaeologists at Xultun in Guatemala identify the earliest named Maya scientist-astronomer, Saktahnwaax, dating to around 800 CE, based on an inscription praising him and a newly deciphered 2,920-day calendar formula showing Maya tracking solar and Venus cycles, evidence of sophisticated scientific inquiry in pre-Columbian Americas.
Physicists Parisi and Zamponi credit Claude, an AI model, with sparking a simple idea that, despite initial errors, guided a clean proof that two parameters in the jamming (granular) model sum to one. The researchers refined Claude’s premise and published their result in the Journal of Statistical Mechanics: Theory and Experiment, illustrating AI’s potential to surface patterns and literature while underscoring that human verification remains essential.
Eight decades after Erdős introduced the probabilistic method, researchers Ma, Shen, and Xie used the geometry of high-dimensional spheres to improve lower bounds on near-diagonal Ramsey numbers, reviving progress on a stubborn problem and signaling a new geometric twist on Erdős’s randomness-based approach.
New findings from crow research by Andreas Nieder show that carrion crows understand zero as a numerical quantity and can perform probabilistic reasoning to pick the more rewarding option, suggesting that the cognitive building blocks of mathematics—quantity sense and probabilistic inference—are ancient and shared with primates and preverbal human infants, implying human math evolved by building on these prehuman foundations.
AI augments, not replaces, human creativity in math and physics by helping formalize informal arguments, check proofs, generate conjectures, and surface overlooked connections. With fast, digital mathematics data and cheap experiments, researchers outline a multi-stage pipeline (setting the agenda, formalizing ideas, proposing conjectures, solving) and point to examples like Aristotle and Axiom Math as progress, while stressing that human insight remains essential.
Mathematicians, led by a Leiden University group and endorsed by the International Mathematical Union, warn that AI’s growing role risks undermining math research through unreliable proofs, poor citation practices, and corporate influence. The Leiden Declaration urges transparency, proper credit, ethical partnerships, and policy actions to protect scholarly integrity as AI disrupts hiring, funding, and dissemination of results—highlighting OpenAI’s recent claim as a case study in rushed, market-driven announcements.