← All stories
● Covered by 8 sources · 55 reportsLow impact24 negative29 neutral2 positive

OpenAI's new model solves ten math problems, prompting concerns about AI's role in mathematics

🔄 Updated 16d ago — new reporting from Ars Technica
New to BrevFeed? We gather this story from every outlet covering it into one summary — ranked by real-world impact, not just the latest headline — so you never miss what matters. What is BrevFeed? →

Key points

  • OpenAI's unreleased model solved ten open math problems.
  • The model's capabilities include solving a coding theory problem.
  • The author discusses potential human responses to AI's mathematical advancements.
  • The article suggests AI could surpass human intuition and teaching in mathematics.
  • AI's influence is linked to a decline in activity on MathOverflow.
  • The workshop on the future of mathematics was held at OpenAI.
  • Sebastian Bubeck asked the author to discuss human mathematical disempowerment.
  • Jacob Tsimerman advised prioritizing detail over correctness.
  • The number of combinatorics papers on arXiv has risen since late 2021.
  • AI's advantage in math problem-solving is its vast symbolic working memory.
  • AI's memory allows it to retain problem details, intermediate steps, and approaches.
  • Human mathematicians can hold only a small number of unfamiliar elements in mind simultaneously.
  • OpenAI published solutions to long-standing mathematical problems.
  • The Verge's Robert Hart is a London-based AI reporter.
  • AI systems are still bad at elementary school arithmetic.
  • AI agents discovered novel mathematical results in an open-world multi-agent environment called "the Station".
  • The Station environment allows AI agents from different model families to pursue a shared research goal.
  • AI agents discovered a new infinite family of finite-field Kakeya sets.
  • AI agents found new exact 604-point kissing configurations in dimension 11.
  • AI agents achieved new records for the discretized Kakeya needle and sign uncertainty problems.
  • AI agents substantially improved the lower bound for Erdős's minimum-overlap problem.
  • AI agents discovered novel infinite families for Book Ramsey numbers.
  • The AI agents produced theorems and analyses explaining how their constructions work.
  • Wang & Wu's preprint proves the Spherical Hadwiger Conjecture.
  • The Spherical Hadwiger Conjecture was open since about 1974.
  • The preprint used OpenAI Codex for assistance.
  • Claude formalized the proof of Fermat’s Last Theorem in Lean.
  • Caltech will host the first hackathon for research-level mathematics.
  • The hackathon will run from October 30th to November 1st.
  • 100 teams will use frontier AI models at the hackathon.
  • The hackathon will last 40 hours.
  • Over $2 million in AI credits will be provided.
  • Erdos's planar unit-distance conjecture was disproved on May 20, 2026.
  • The planar unit-distance conjecture was open for 80 years.
  • The first explicit non-sofic group was constructed on August 1, 2026.
  • The non-sofic group problem was open for 27 years.
  • The six-sphere was shown to admit a complex structure on August 23, 2026.
  • The six-sphere problem was open for 78 years.
  • Leading mathematicians will assess the teams' understanding of results.
  • NYU professor Tristan Buckmaster claims OpenAI used his team's progress on the Navier-Stokes problem.
  • Buckmaster announced three proofs with a preliminary finding on the Navier-Stokes problem.
  • Buckmaster collaborated with Anthropic mathematician Levent Alpöge.
  • Buckmaster's team used both Codex and Claude AI models.
  • OpenAI's head of mathematical research, Sébastien Bubeck, denies Buckmaster's claims.
  • OpenAI announced solving the Navier-Stokes problem on a Tuesday.
  • OpenAI's internal AI model is more powerful than GPT-6 Astra.
  • OpenAI's internal AI model used 10,000 concurrent agents.
  • The Navier-Stokes problem is one of seven Millennium Prize Problems.
  • OpenAI started training its internal AI model on August 28th.
  • OpenAI's AI model solved a significant aspect of the Navier-Stokes equations.
  • OpenAI's AI model solved the Navier-Stokes problem in 88 hours.
  • The Clay Mathematics Institute runs the Millennium Prize.
  • Terence Tao raised concerns that AI is consuming open mathematical problems at an unsustainable rate.
  • The Navier-Stokes problem has stumped humans for nearly a century.
  • The Clay Mathematics Institute offers a $1 million reward for solutions to each Millennium Prize Problem.
  • Only one other Millennium Prize Problem has been solved since 2000.
  • The Navier-Stokes problem centers on fluid dynamics.
  • OpenAI's unreleased model resolved the Navier-Stokes existence and smoothness problem.
  • The Millennium Prize Problems have offered a $1,000,000 prize since May 24th, 2000.
  • Tristan Buckmaster collaborated with Levent Alpöge.
  • Buckmaster and Alpöge worked on the problem for almost a year.
  • Buckmaster and Alpöge used GPT-5.6 Sol.
  • Buckmaster and Alpöge had a breakthrough on August 15th.
  • OpenAI's internal AI model used coordinating agents.
  • The agents could read from a cached internet version and run code.
  • Agents were subdivided into groups that could communicate.
  • The group that solved Navier-Stokes involved 10,000 concurrent agents.
  • The agents arrived at their resolution on Saturday, September 5.
  • The Clay Mathematics Institute has not commented on OpenAI's solution.
  • OpenAI's solution to Navier-Stokes is met with plagiarism controversy.
  • Tristan Buckmaster alleges OpenAI plagiarized his team's work on Euler's equations.
  • Tristan Buckmaster claims OpenAI issued career threats.
  • Euler's equations are a stepping stone to solving Navier-Stokes.
  • Buckmaster and Alpöge's work was a personal collaboration, free of institutional agreements.
  • The author's mental model of AI broke on September 8, 2026.
  • OpenAI's announcement was on a Tuesday.
  • Abhishek Saha is a mathematics professor at Queen Mary University of London.
  • Mathematician Andreas Thom accused OpenAI of dishonesty and lack of transparency.
  • Thom raised concerns that his interactions with ChatGPT contributed to OpenAI's mathematical breakthroughs.
  • One of OpenAI's 10 results involved Thom's expertise in non-sofic groups.
  • OpenAI acknowledged its result built on previous work by Thom and Gábor Kun.
  • OpenAI's Navier-Stokes proof included a Lean 4 formal proof.
  • Formal proofs for other AI-settled conjectures also used Lean 4.
  • Formalizing one page of undergraduate math took 40 hours in 2005.
  • The mathematical community views AI companies' focus on solving complex problems as detrimental to mathematics.
  • AI companies' goals are misaligned with the mathematical community's goals.
  • Twenty-five Fields Medal-winning mathematicians signed an open letter criticizing AI labs.
  • OpenAI withdrew its sponsorship of a Caltech math event.
  • AI-generated solutions must be understandable and integrated into the mathematical community.
  • The Clay Mathematics Institute announced the Navier-Stokes problem was settled on September 11, 2026.
  • The CMI unveiled the Millennium Prize Problems at a meeting in Paris in 2000.
  • OpenAI's solution to a Millennium Prize Problem cost an estimated $15 million.
  • Colva Roney-Dougal, head of pure mathematics at the University of St Andrews, expressed shock at the AI's rapid progress.
  • The Fields medalists' letter is titled "A Severe Misalignment of AI in Mathematics".
  • Po-Shen Loh wrote the guest post.
  • Claude Code generated the webpage design, layout, headings, and summaries.
  • The Leiden Declaration has over 4,000 signatories.
  • The Math and AI declaration has over 7,000 signatories.
  • The open letter opposing the Caltech Mathathon has over 2,000 signatories.
  • The Advisory Group on Mathematics and Artificial Intelligence (AGMAI) was established.
  • AGMAI is hosted at the Institute for Advanced Study (Princeton).
  • AGMAI's website is agmai.org.
  • AGMAI advises AI companies on interactions with mathematical research and community.
  • AGMAI ensures responsible presentation and release of AI-generated mathematical results.
  • AGMAI serves as a communication channel between mathematicians and the AI industry.
  • AGMAI operates independently of any AI company.
  • OpenAI's internal model has resolved over 100 additional open problems.
  • OpenAI announced the new advisory group on a Monday.
  • AGMAI will advise on the review and communication of emerging results.
  • The AGMAI group has nine members.
  • The AGMAI members are prominent researchers in the field.
  • Amit Sahai wrote the guest post.
  • AGMAI's launch was described as disorganized by mathematicians.
  • AGMAI's first task is coordinating the release of numerous new mathematical results from an unreleased OpenAI model.
  • New recommendations address the responsible release of AI-generated mathematical results.
  • Guidelines emphasize human understanding, verification, and adherence to established mathematical scholarly norms.
  • The recommendations were published on September 29, 2026.
  • Frontier AI labs are testing advanced mathematical problems on proprietary models.
  • The recommendations ask AI labs to stop testing advanced mathematical problems on proprietary models.
  • A key scholarly norm is that authors should understand, verify, and take responsibility for their paper's content.
  • Authors of significant mathematical advancements regularly give seminars and talks to explain their work.
  • OpenAI established an independent advisory group of mathematicians.
  • The advisory group aims to guide OpenAI's interactions with the mathematical community.
  • The advisory group aims to improve how AI research results are presented and released.
  • Erdosproblems.com is undergoing changes due to AI's impact on solving mathematical problems.
  • Erdosproblems.com was created to promote problems to a human audience.
  • Erdosproblems.com became a benchmark for AI capabilities.
  • AI's progress in mathematics has led to solving many previously unanswered questions.
  • Increased human efforts and renewed interest contributed to recent progress in solving problems.
  • Some mathematicians dismiss Erdős-style mathematics as 'easy/recreational'.
  • OpenAI released 722 manuscripts detailing solutions to mathematical problems.
  • The manuscripts cover 372 result families.
  • The release follows calls from AGMAI for prompt and transparent publication of AI-generated mathematical results.
  • OpenAI released 722 manuscripts on GitHub.
  • The release details AI-generated solutions and progress for 372 major mathematical problems.
  • The release stems from an unreleased ChatGPT pioneer model.
  • The average result required around three hours of ChatGPT Pro use.
  • OpenAI included the model's reasoning, compute estimates, and number of problems attempted.
  • OpenAI followed AGMAI guidelines for the release of the papers.
  • Guidelines included details like the model name, prompts, and compute costs.
  • AI autoformalization does not guarantee correctness of original natural language arguments.
  • Resolving ambiguities in mathematical natural language text is arbitrarily high in the Solvability Complexity Index.
  • The issue was observed in OpenAI's Navier-Stokes proof.
  • The formal Lean proof did not correspond to the natural language proof.
  • OpenAI published over 370 mathematical results across algebra, theoretical computer science, and mathematical logic.
  • The Institute for Advanced Study in Princeton does not endorse testing advanced mathematical problems on proprietary AI models.
  • AI can output mathematical arguments without human understanding, verification, or responsibility.
  • OpenAI released a proof of Subhash Khot's Unique Games Conjecture (UGC).
  • The UGC implies many optimization problems are NP-hard.
  • Timothy Gowers and Edward Witten are part of OpenAI's advisory group.
  • OpenAI's math solutions do not yet meet field standards.
  • OpenAI's advisory group stated OpenAI did not fully meet their guidelines.
  • A new paper highlighted gaps between natural language and formally expressed solutions.
  • AGMAI is hosted by Princeton University's Institute for Advanced Studies.
  • OpenAI claimed the Partition Principle does not imply the Axiom of Choice.
  • A mathematician criticized OpenAI's preprint for issues with clarity, terminology, and unusual theorem proofs.
  • The mathematician noted OpenAI's paper referenced unpublished and unrefereed works.
  • The mathematician's own paper was inappropriately referenced in OpenAI's preprint.
  • OpenAI released nearly 400 AI-generated results.
  • The results cover diverse mathematical disciplines, including combinatorics, geometry, number theory, and theoretical computer science.
  • OpenAI announced solving numerous open mathematical problems on October 6.
  • OpenAI's solutions include complex issues in group theory and Quasi-Riemann problems.
  • The announcement prompted discussion among mathematicians about the future of mathematical research and AI's role.
  • The Hacker News Front Page post was updated on October 10, 2026, at 10:39 am (CEST).
  • Tim Santens commented on the Quasi-Riemann result.
  • Only 300 of OpenAI's 722 released solutions were formalized.
  • Mathematicians are concerned about the validity and utility of OpenAI's AI-generated results.
  • Critics argue that the lack of transparency regarding the AI model and methods places an undue burden on the mathematical community.
  • The volume of unverified solutions could hinder genuine mathematical progress.
  • Researchers say it could take years for humans to validate and prove all the AI-driven work.
  • Mathematicians argue that AI-powered research and release threaten the field itself.
  • There is little benefit to solutions that leap from a prompt to a solution without clear working.
  • Mathematicians may not be able to build on AI results without knowing how they were achieved.
  • The development of AI solutions has not produced new fields of mathematical understanding.
  • The confluence of mathematical and cybersecurity capabilities of frontier AI could question cryptography's sanctity.
  • AI's capabilities could put trillions of dollars worth of cryptocurrency at risk.
  • A community project, "oai-math-revised," was launched to revise OpenAI's math write-ups.
  • The project aims to improve the clarity and citation completeness of OpenAI's math write-ups.
  • The project invites participation from the broader math community.
  • The project's GitHub repository is https://github.com/mathcommunity/oai-math-revised/.
  • The project's browsable repository is https://mathcommunity.github.io/oai-math-revised/.
  • Abhinav Nellore, Johannes Schmitt, Soledad Villar, and Rachel Ward wrote the guest post.
  • The project seeks admins/reviewers for categories listed in openai/math/blob/main/overview.pdf.
  • Richard Borcherds is a UC Berkeley math professor.
  • Richard Borcherds is a Fields medalist.
  • Richard Borcherds stopped grading homework for correctness.
  • Richard Borcherds now grades homework on completion.
  • Richard Borcherds asked ChatGPT for career advice regarding AI.
  • OpenAI announced three new scientific discoveries.
  • The discoveries are in mathematics, computer science, and physics.
  • OpenAI is using external experts as "shepherds" for some discoveries.
  • OpenAI is releasing raw AI output for some discoveries.
  • OpenAI's math advisory panel raised concerns about release practices.
  • OpenAI's new math discovery is "Local smoothing for the Euclidean wave equation."
  • The math proof uses an auxiliary partial differential equation.
  • The math proof establishes Sogge’s local smoothing for waves in every spatial dimension at least two.
  • The math proof yields the Kakeya maximal and dimension conjectures.
  • The math proof yields Stein’s spherical restriction estimates.
  • OpenAI formed groups of experts to examine and explain results.
  • OpenAI published two new manuscripts on GitHub.
  • OpenAI is welcoming feedback on broadening participation and making the process more useful.
  • OpenAI's solutions include a refutation of the Exponential Time Hypothesis (ETH).
  • Anthropic also released its own refutation of ETH.
  • OpenAI's release process for math involved GitHub, computer scientists formed a working group, and physicists formed a working group.
  • MIT professor Ryan Williams is organizing the computer science field's response to the ETH refutation.
  • Richard Borcherds now primarily uses in-person exams with no electronic devices.
  • Richard Borcherds noted online exam scores increased significantly due to AI use.
  • Richard Borcherds stated grading for correctness incentivizes AI use and penalizes honest students.
  • Frank Calegari wrote the guest post.
  • The guest post was converted using AI.
  • OpenAI's papers are atrociously written and violate community expectations of attribution.
  • Some of OpenAI's papers are wrong.
  • Jessica Hullman is the Ginni Rometty Professor of Computer Science and Policy Fellow at Northwestern University.
  • Bill Thurston wrote an essay on the role of rigorous proofs in mathematical advances.
  • Axiom Math formally verified the Stacks Project.
  • The Stacks Project has 7,654 pages translated into 16 million lines of code.
  • Axiom Math is a $1.6 billion startup.
  • Johan de Jong started the Stacks Project in 2005.

AI Solves Open Math Problems

OpenAI recently announced that an unreleased AI model successfully solved ten open problems in mathematics. One of these problems was in coding theory, which is noted as a significant achievement. This event marks a notable advancement in AI's capability within complex mathematical domains.

Implications for Human Mathematicians

The author explores various ways humans might react to AI's increasing proficiency in mathematics. These reactions, termed 'copes,' include believing humans will direct AI, teach AI-discovered mathematics, or canonize AI's results. The article suggests these coping mechanisms may be refuted by reality as AI capabilities grow.

AI's Potential Autonomy in Research

The article posits that AI could eventually exceed human intuition and taste in mathematical research, leading to scenarios where AI independently identifies and solves problems more effectively than human-guided approaches. It also suggests AI could become superior teachers of mathematics and architects of mathematical structures.

Future of Mathematical Understanding

A key concern raised is that AI could delve so deeply into advanced mathematics that the knowledge gap between elementary concepts and the frontier becomes too vast for humans to bridge. The author suggests that in such a future, AI would perform frontier math, science, and engineering, with no human understanding required, potentially outcompeting firms that insist on human oversight.

Updates

🕒 2026-11-05 · new reporting from TechCrunch
  • Axiom Math formally verified the Stacks Project.
  • The Stacks Project has 7,654 pages translated into 16 million lines of code.
  • Axiom Math is a $1.6 billion startup.
  • Johan de Jong started the Stacks Project in 2005.
🕒 2026-11-04 · new reporting from Hacker News Front Page
  • Jessica Hullman is the Ginni Rometty Professor of Computer Science and Policy Fellow at Northwestern University.
  • Bill Thurston wrote an essay on the role of rigorous proofs in mathematical advances.
🕒 2026-11-01 · new reporting from Hacker News Front Page
  • Frank Calegari wrote the guest post.
  • The guest post was converted using AI.
  • OpenAI's papers are atrociously written and violate community expectations of attribution.
  • Some of OpenAI's papers are wrong.
🕒 2026-10-27 · new reporting from TechCrunch, Tom's Hardware, The Verge
  • OpenAI's solutions include a refutation of the Exponential Time Hypothesis (ETH).
  • Anthropic also released its own refutation of ETH.
  • OpenAI's release process for math involved GitHub, computer scientists formed a working group, and physicists formed a working group.
  • MIT professor Ryan Williams is organizing the computer science field's response to the ETH refutation.
  • Richard Borcherds now primarily uses in-person exams with no electronic devices.
  • Richard Borcherds noted online exam scores increased significantly due to AI use.
  • Richard Borcherds stated grading for correctness incentivizes AI use and penalizes honest students.
🕒 2026-10-27 · new reporting from The Verge, Engadget
  • OpenAI announced three new scientific discoveries.
  • The discoveries are in mathematics, computer science, and physics.
  • OpenAI is using external experts as "shepherds" for some discoveries.
  • OpenAI is releasing raw AI output for some discoveries.
  • OpenAI's math advisory panel raised concerns about release practices.
  • OpenAI's new math discovery is "Local smoothing for the Euclidean wave equation."
  • The math proof uses an auxiliary partial differential equation.
  • The math proof establishes Sogge’s local smoothing for waves in every spatial dimension at least two.
  • The math proof yields the Kakeya maximal and dimension conjectures.
  • The math proof yields Stein’s spherical restriction estimates.
  • OpenAI formed groups of experts to examine and explain results.
  • OpenAI published two new manuscripts on GitHub.
  • OpenAI is welcoming feedback on broadening participation and making the process more useful.
🕒 2026-10-25 · new reporting from Hacker News Front Page
  • Richard Borcherds is a UC Berkeley math professor.
  • Richard Borcherds is a Fields medalist.
  • Richard Borcherds stopped grading homework for correctness.
  • Richard Borcherds now grades homework on completion.
  • Richard Borcherds asked ChatGPT for career advice regarding AI.
🕒 2026-10-24 · new reporting from Hacker News Front Page
  • A community project, "oai-math-revised," was launched to revise OpenAI's math write-ups.
  • The project aims to improve the clarity and citation completeness of OpenAI's math write-ups.
  • The project invites participation from the broader math community.
  • The project's GitHub repository is https://github.com/mathcommunity/oai-math-revised/.
  • The project's browsable repository is https://mathcommunity.github.io/oai-math-revised/.
  • Abhinav Nellore, Johannes Schmitt, Soledad Villar, and Rachel Ward wrote the guest post.
  • The project seeks admins/reviewers for categories listed in openai/math/blob/main/overview.pdf.
🕒 2026-10-13 · new reporting from Tom's Hardware
  • Only 300 of OpenAI's 722 released solutions were formalized.
  • Mathematicians are concerned about the validity and utility of OpenAI's AI-generated results.
  • Critics argue that the lack of transparency regarding the AI model and methods places an undue burden on the mathematical community.
  • The volume of unverified solutions could hinder genuine mathematical progress.
  • Researchers say it could take years for humans to validate and prove all the AI-driven work.
  • Mathematicians argue that AI-powered research and release threaten the field itself.
  • There is little benefit to solutions that leap from a prompt to a solution without clear working.
  • Mathematicians may not be able to build on AI results without knowing how they were achieved.
  • The development of AI solutions has not produced new fields of mathematical understanding.
  • The confluence of mathematical and cybersecurity capabilities of frontier AI could question cryptography's sanctity.
  • AI's capabilities could put trillions of dollars worth of cryptocurrency at risk.
🕒 2026-10-10 · new reporting from Hacker News Front Page
  • OpenAI announced solving numerous open mathematical problems on October 6.
  • OpenAI's solutions include complex issues in group theory and Quasi-Riemann problems.
  • The announcement prompted discussion among mathematicians about the future of mathematical research and AI's role.
  • The Hacker News Front Page post was updated on October 10, 2026, at 10:39 am (CEST).
  • Tim Santens commented on the Quasi-Riemann result.
🕒 2026-10-09 · new reporting from The Verge
  • OpenAI released nearly 400 AI-generated results.
  • The results cover diverse mathematical disciplines, including combinatorics, geometry, number theory, and theoretical computer science.
🕒 2026-10-09 · new reporting from Hacker News Front Page
  • OpenAI claimed the Partition Principle does not imply the Axiom of Choice.
  • A mathematician criticized OpenAI's preprint for issues with clarity, terminology, and unusual theorem proofs.
  • The mathematician noted OpenAI's paper referenced unpublished and unrefereed works.
  • The mathematician's own paper was inappropriately referenced in OpenAI's preprint.
🕒 2026-10-08 · new reporting from TechCrunch
  • OpenAI's math solutions do not yet meet field standards.
  • OpenAI's advisory group stated OpenAI did not fully meet their guidelines.
  • A new paper highlighted gaps between natural language and formally expressed solutions.
  • AGMAI is hosted by Princeton University's Institute for Advanced Studies.
🕒 2026-10-07 · new reporting from Engadget, Hacker News Front Page, Guardian Technology
  • OpenAI released 722 manuscripts on GitHub.
  • The release details AI-generated solutions and progress for 372 major mathematical problems.
  • The release stems from an unreleased ChatGPT pioneer model.
  • The average result required around three hours of ChatGPT Pro use.
  • OpenAI included the model's reasoning, compute estimates, and number of problems attempted.
  • OpenAI followed AGMAI guidelines for the release of the papers.
  • Guidelines included details like the model name, prompts, and compute costs.
  • AI autoformalization does not guarantee correctness of original natural language arguments.
  • Resolving ambiguities in mathematical natural language text is arbitrarily high in the Solvability Complexity Index.
  • The issue was observed in OpenAI's Navier-Stokes proof.
  • The formal Lean proof did not correspond to the natural language proof.
  • OpenAI published over 370 mathematical results across algebra, theoretical computer science, and mathematical logic.
  • The Institute for Advanced Study in Princeton does not endorse testing advanced mathematical problems on proprietary AI models.
  • AI can output mathematical arguments without human understanding, verification, or responsibility.
  • OpenAI released a proof of Subhash Khot's Unique Games Conjecture (UGC).
  • The UGC implies many optimization problems are NP-hard.
  • Timothy Gowers and Edward Witten are part of OpenAI's advisory group.
🕒 2026-10-07 · new reporting from The Verge
  • OpenAI released 722 manuscripts detailing solutions to mathematical problems.
  • The manuscripts cover 372 result families.
  • The release follows calls from AGMAI for prompt and transparent publication of AI-generated mathematical results.
🕒 2026-10-06 · new reporting from Hacker News Front Page
  • Erdosproblems.com is undergoing changes due to AI's impact on solving mathematical problems.
  • Erdosproblems.com was created to promote problems to a human audience.
  • Erdosproblems.com became a benchmark for AI capabilities.
  • AI's progress in mathematics has led to solving many previously unanswered questions.
  • Increased human efforts and renewed interest contributed to recent progress in solving problems.
  • Some mathematicians dismiss Erdős-style mathematics as 'easy/recreational'.
🕒 2026-10-05 · new reporting from The Verge
  • OpenAI established an independent advisory group of mathematicians.
  • The advisory group aims to guide OpenAI's interactions with the mathematical community.
  • The advisory group aims to improve how AI research results are presented and released.
🕒 2026-09-30 · new reporting from Hacker News Front Page
  • New recommendations address the responsible release of AI-generated mathematical results.
  • Guidelines emphasize human understanding, verification, and adherence to established mathematical scholarly norms.
  • The recommendations were published on September 29, 2026.
  • Frontier AI labs are testing advanced mathematical problems on proprietary models.
  • The recommendations ask AI labs to stop testing advanced mathematical problems on proprietary models.
  • A key scholarly norm is that authors should understand, verify, and take responsibility for their paper's content.
  • Authors of significant mathematical advancements regularly give seminars and talks to explain their work.
🕒 2026-09-28 · new reporting from The Verge
  • AGMAI's launch was described as disorganized by mathematicians.
  • AGMAI's first task is coordinating the release of numerous new mathematical results from an unreleased OpenAI model.
🕒 2026-09-26 · new reporting from Hacker News Front Page
  • Amit Sahai wrote the guest post.
🕒 2026-09-23 · new reporting from The Verge
  • AGMAI will advise on the review and communication of emerging results.
  • The AGMAI group has nine members.
  • The AGMAI members are prominent researchers in the field.
🕒 2026-09-21 · new reporting from Hacker News Front Page, TechCrunch
  • The Advisory Group on Mathematics and Artificial Intelligence (AGMAI) was established.
  • AGMAI is hosted at the Institute for Advanced Study (Princeton).
  • AGMAI's website is agmai.org.
  • AGMAI advises AI companies on interactions with mathematical research and community.
  • AGMAI ensures responsible presentation and release of AI-generated mathematical results.
  • AGMAI serves as a communication channel between mathematicians and the AI industry.
  • AGMAI operates independently of any AI company.
  • OpenAI's internal model has resolved over 100 additional open problems.
  • OpenAI announced the new advisory group on a Monday.
🕒 2026-09-20 · new reporting from Hacker News Front Page
  • Po-Shen Loh wrote the guest post.
  • Claude Code generated the webpage design, layout, headings, and summaries.
  • The Leiden Declaration has over 4,000 signatories.
  • The Math and AI declaration has over 7,000 signatories.
  • The open letter opposing the Caltech Mathathon has over 2,000 signatories.
🕒 2026-09-13 · new reporting from Hacker News Front Page
  • The Fields medalists' letter is titled "A Severe Misalignment of AI in Mathematics".
🕒 2026-09-12 · new reporting from Guardian Technology
  • OpenAI's solution to a Millennium Prize Problem cost an estimated $15 million.
  • Colva Roney-Dougal, head of pure mathematics at the University of St Andrews, expressed shock at the AI's rapid progress.
🕒 2026-09-12 · new reporting from Hacker News Front Page
  • The Clay Mathematics Institute announced the Navier-Stokes problem was settled on September 11, 2026.
  • The CMI unveiled the Millennium Prize Problems at a meeting in Paris in 2000.
🕒 2026-09-11 · new reporting from Hacker News Front Page, TechCrunch
  • The mathematical community views AI companies' focus on solving complex problems as detrimental to mathematics.
  • AI companies' goals are misaligned with the mathematical community's goals.
  • Twenty-five Fields Medal-winning mathematicians signed an open letter criticizing AI labs.
  • OpenAI withdrew its sponsorship of a Caltech math event.
  • AI-generated solutions must be understandable and integrated into the mathematical community.
🕒 2026-09-11 · new reporting from Hacker News Front Page
  • OpenAI's Navier-Stokes proof included a Lean 4 formal proof.
  • Formal proofs for other AI-settled conjectures also used Lean 4.
  • Formalizing one page of undergraduate math took 40 hours in 2005.
🕒 2026-09-10 · new reporting from The Verge
  • Mathematician Andreas Thom accused OpenAI of dishonesty and lack of transparency.
  • Thom raised concerns that his interactions with ChatGPT contributed to OpenAI's mathematical breakthroughs.
  • One of OpenAI's 10 results involved Thom's expertise in non-sofic groups.
  • OpenAI acknowledged its result built on previous work by Thom and Gábor Kun.
🕒 2026-09-09 · new reporting from The Verge
  • OpenAI's announcement was on a Tuesday.
  • Abhishek Saha is a mathematics professor at Queen Mary University of London.
🕒 2026-09-09 · new reporting from Hacker News Front Page
  • The author's mental model of AI broke on September 8, 2026.
🕒 2026-09-09 · new reporting from Tom's Hardware
  • OpenAI's solution to Navier-Stokes is met with plagiarism controversy.
  • Tristan Buckmaster alleges OpenAI plagiarized his team's work on Euler's equations.
  • Tristan Buckmaster claims OpenAI issued career threats.
  • Euler's equations are a stepping stone to solving Navier-Stokes.
  • Buckmaster and Alpöge's work was a personal collaboration, free of institutional agreements.
🕒 2026-09-09 · new reporting from CNBC Technology
  • OpenAI's internal AI model used coordinating agents.
  • The agents could read from a cached internet version and run code.
  • Agents were subdivided into groups that could communicate.
  • The group that solved Navier-Stokes involved 10,000 concurrent agents.
  • The agents arrived at their resolution on Saturday, September 5.
  • The Clay Mathematics Institute has not commented on OpenAI's solution.
🕒 2026-09-09 · new reporting from Hacker News Front Page
  • OpenAI's unreleased model resolved the Navier-Stokes existence and smoothness problem.
  • The Millennium Prize Problems have offered a $1,000,000 prize since May 24th, 2000.
  • Tristan Buckmaster collaborated with Levent Alpöge.
  • Buckmaster and Alpöge worked on the problem for almost a year.
  • Buckmaster and Alpöge used GPT-5.6 Sol.
  • Buckmaster and Alpöge had a breakthrough on August 15th.
🕒 2026-09-09 · new reporting from Hacker News Front Page, Guardian Technology, Engadget
  • Terence Tao raised concerns that AI is consuming open mathematical problems at an unsustainable rate.
  • The Navier-Stokes problem has stumped humans for nearly a century.
  • The Clay Mathematics Institute offers a $1 million reward for solutions to each Millennium Prize Problem.
  • Only one other Millennium Prize Problem has been solved since 2000.
  • The Navier-Stokes problem centers on fluid dynamics.
🕒 2026-09-08 · new reporting from The Verge, BBC Technology
  • OpenAI announced solving the Navier-Stokes problem on a Tuesday.
  • OpenAI's internal AI model is more powerful than GPT-6 Astra.
  • OpenAI's internal AI model used 10,000 concurrent agents.
  • The Navier-Stokes problem is one of seven Millennium Prize Problems.
  • OpenAI started training its internal AI model on August 28th.
  • OpenAI's AI model solved a significant aspect of the Navier-Stokes equations.
  • OpenAI's AI model solved the Navier-Stokes problem in 88 hours.
  • The Clay Mathematics Institute runs the Millennium Prize.
🕒 2026-09-08 · new reporting from TechCrunch
  • NYU professor Tristan Buckmaster claims OpenAI used his team's progress on the Navier-Stokes problem.
  • Buckmaster announced three proofs with a preliminary finding on the Navier-Stokes problem.
  • Buckmaster collaborated with Anthropic mathematician Levent Alpöge.
  • Buckmaster's team used both Codex and Claude AI models.
  • OpenAI's head of mathematical research, Sébastien Bubeck, denies Buckmaster's claims.
🕒 2026-09-07 · new reporting from Hacker News Front Page
  • Caltech will host the first hackathon for research-level mathematics.
  • The hackathon will run from October 30th to November 1st.
  • 100 teams will use frontier AI models at the hackathon.
  • The hackathon will last 40 hours.
  • Over $2 million in AI credits will be provided.
  • Erdos's planar unit-distance conjecture was disproved on May 20, 2026.
  • The planar unit-distance conjecture was open for 80 years.
  • The first explicit non-sofic group was constructed on August 1, 2026.
  • The non-sofic group problem was open for 27 years.
  • The six-sphere was shown to admit a complex structure on August 23, 2026.
  • The six-sphere problem was open for 78 years.
  • Leading mathematicians will assess the teams' understanding of results.
🕒 2026-09-07 · new reporting from Hacker News Front Page
  • Wang & Wu's preprint proves the Spherical Hadwiger Conjecture.
  • The Spherical Hadwiger Conjecture was open since about 1974.
  • The preprint used OpenAI Codex for assistance.
  • Claude formalized the proof of Fermat’s Last Theorem in Lean.
🕒 2026-08-28 · new reporting from Hacker News Front Page
  • AI agents discovered novel mathematical results in an open-world multi-agent environment called "the Station".
  • The Station environment allows AI agents from different model families to pursue a shared research goal.
  • AI agents discovered a new infinite family of finite-field Kakeya sets.
  • AI agents found new exact 604-point kissing configurations in dimension 11.
  • AI agents achieved new records for the discretized Kakeya needle and sign uncertainty problems.
  • AI agents substantially improved the lower bound for Erdős's minimum-overlap problem.
  • AI agents discovered novel infinite families for Book Ramsey numbers.
  • The AI agents produced theorems and analyses explaining how their constructions work.
🕒 2026-08-20 · new reporting from The Verge
  • OpenAI published solutions to long-standing mathematical problems.
  • The Verge's Robert Hart is a London-based AI reporter.
  • AI systems are still bad at elementary school arithmetic.
🕒 2026-08-15 · new reporting from Hacker News Front Page
  • AI's advantage in math problem-solving is its vast symbolic working memory.
  • AI's memory allows it to retain problem details, intermediate steps, and approaches.
  • Human mathematicians can hold only a small number of unfamiliar elements in mind simultaneously.
🕒 2026-08-15 · new reporting from Hacker News Front Page
  • AI's influence is linked to a decline in activity on MathOverflow.
  • The workshop on the future of mathematics was held at OpenAI.
  • Sebastian Bubeck asked the author to discuss human mathematical disempowerment.
  • Jacob Tsimerman advised prioritizing detail over correctness.
  • The number of combinatorics papers on arXiv has risen since late 2021.

✨ This summary was generated by AI from the outlets' reporting listed below. It is not independently verified and may contain errors — check the original sources. How BrevFeed works →

The daily brief

One email each morning: the day's tech stories, clustered across outlets and summarized. No account needed.

One email a day. Unsubscribe in one click, any time.

Today's brief

Spend a few minutes, get the whole day. Every topic's top stories in one hands-free rundown — listen, watch, or read the transcript.

~4 min · 3 stories · Nov 21

▶ Play today's brief Listen on Spotify

New every morning, and the back catalogue is archived by date.

How outlets covered it

Axiom Math announced it has formally verified the Stacks Project, a large body of human-written mathematics, by translating its 7,654 pages into 16 million lines of computer-checked code. This initiative aims to provide a verified mathematical foundation for AI agents to prevent errors and ensure reliability in AI-generated software and scientific work.

This essay argues that science requires human understanding, not just verifiable AI-generated answers. It discusses the tension between prioritizing outputs from AI and the human insight necessary for scientific progress and integration of new findings.

A mathematician reflects on the implications of OpenAI's release of over 700 research papers, which demonstrated AI's capability in mathematics. The author addresses concerns within the mathematical community regarding AI's impact and the quality of some AI-generated research.

OpenAI announced new breakthroughs in mathematics, computer science, and physics, alongside an updated approach to sharing its findings. Researchers are expressing frustration with OpenAI's method of releasing information, particularly after a previous release of numerous preprints.

A UC Berkeley professor has stopped assigning graded homework, citing the difficulty in distinguishing AI-generated submissions from student work. This change reflects a broader challenge in academia regarding AI's impact on assessment methods, leading to a shift towards in-person, proctored exams.

OpenAI released solutions to three math problems, including a refutation of the Exponential Time Hypothesis (ETH), a foundational assumption in computer science. Anthropic also released its own refutation of ETH shortly after. This development challenges long-held understandings of computational difficulty and has prompted discussion within the scientific community.

OpenAI released new discoveries in mathematics, computer science, and physics, following previous math-only releases. The company is using external expert groups to examine and explain these AI-generated results, including a potential disproof of the Exponential Time Hypothesis.

OpenAI announced three new scientific discoveries in mathematics, computer science, and physics, following its previous release of mathematical solutions. The company is adopting a new approach to academic collaboration, using external experts as "shepherds" for some discoveries and releasing raw AI output for others, while also addressing concerns from its math advisory panel regarding release practices.

A UC Berkeley math professor, Richard Borcherds, has stopped grading homework for correctness because students are using AI to achieve top scores. This change reflects the increasing impact of AI on academic integrity and the challenges it poses for traditional assessment methods in education.

A new community project, "oai-math-revised," has been launched to revise and improve the clarity and citation completeness of OpenAI's math write-ups available on GitHub. The project invites participation from the broader math community to enhance these educational resources.

OpenAI released over 722 AI-generated solutions to mathematical problems, with only 300 formalized, leading to concerns from mathematicians about the validity and utility of the results. Critics argue that the lack of transparency regarding the AI model and methods, combined with the volume of unverified solutions, places an undue burden on the mathematical community and could hinder genuine progress.

OpenAI released nearly 400 AI-generated mathematical results across more than 700 manuscripts, covering diverse mathematical disciplines. The volume of the release is significant, and mathematicians anticipate it will take years to fully comprehend the implications and content.

OpenAI announced on October 6 that it had solved numerous open mathematical problems, including complex issues in group theory and Quasi-Riemann problems. This development has prompted significant discussion among mathematicians regarding the future of mathematical research and the role of AI in the field.

A mathematician criticized OpenAI's preprint claiming the Partition Principle does not imply the Axiom of Choice, citing issues with clarity, terminology, and unusual theorem proofs. The author also noted the paper referenced unpublished and unrefereed works, including one of their own papers inappropriately.

OpenAI released hundreds of claimed solutions to complex math problems, but an advisory group of mathematicians (AGMAI) stated that OpenAI did not fully meet their guidelines. Key issues include the use of proprietary models, insufficient formalization of proofs, and a lack of human-understandable explanations for the AI's reasoning, raising concerns about the practical utility of these solutions.

OpenAI, advised by mathematicians like Timothy Gowers and Edward Witten, released 372 AI-generated mathematical proofs, including a proof for Subhash Khot's Unique Games Conjecture (UGC). The UGC proof, which implies many optimization problems are NP-hard, was generated by an AI and is currently being scrutinized by human mathematicians for understanding and validation.

OpenAI released over 370 new mathematical findings across various topics, following its recent solution of the Navier-Stokes equation. This action has raised concerns among mathematicians regarding the use of proprietary AI models for advanced research and the lack of human understanding or verification of the AI-generated arguments.

A new paper demonstrates that AI autoformalization, which translates natural language mathematical texts into formal languages like Lean for mechanical verification, does not guarantee the correctness of the original natural language argument. The difficulty of resolving ambiguities in mathematical natural language text for semantically faithful translation is arbitrarily high in the Solvability Complexity Index, making it harder than any computational problem, including the Halting problem. This issue was observed in OpenAI's announced Navier-Stokes proof, where the formal Lean proof did not correspond to the natural language proof.

OpenAI released 722 manuscripts on GitHub, detailing AI-generated solutions and progress for 372 major mathematical problems. This release, stemming from an unreleased ChatGPT pioneer model, is expected to intensify ongoing discussions within the mathematics community regarding AI's role in solving complex problems.

OpenAI released 722 manuscripts detailing solutions to 372 families of long-standing mathematical problems, generated by an unreleased frontier model. This release follows calls from the Advisory Group on Mathematics and Artificial Intelligence (AGMAI) for prompt and transparent publication of AI-generated mathematical results.

The website Erdosproblems.com is undergoing changes due to the increasing impact of AI on solving mathematical problems. The site, originally intended to promote problems to human mathematicians, has become a benchmark for AI capabilities, leading to both progress in solving problems and concerns about AI-generated solutions displacing human interest.

OpenAI has established an independent advisory group of mathematicians to guide its interactions with the mathematical community, following criticism over its handling of AI breakthroughs in mathematics. This initiative aims to improve how AI research results are presented and released, addressing previous missteps that alienated researchers.

A new set of recommendations addresses the responsible release of AI-generated mathematical results, particularly as some AI labs test advanced problems on proprietary models. The guidelines emphasize human understanding, verification, and adherence to established mathematical scholarly norms. This initiative aims to integrate AI advancements into the mathematical community while maintaining academic integrity and transparency.

OpenAI established the Advisory Group on Mathematics and Artificial Intelligence (AGMAI) to improve relations with the mathematical community and advise on AI math breakthroughs. However, mathematicians, including a group member, describe the launch as disorganized, suggesting OpenAI has not learned from past communication issues. The group's first task involves coordinating the release of numerous new mathematical results from an unreleased OpenAI model, causing concern among researchers.

AI systems are generating new mathematical ideas at a pace that challenges human researchers. This development necessitates a re-evaluation of how the mathematics community engages with and understands these AI-driven discoveries.

OpenAI established a new independent panel of mathematicians, the Advisory Group on Mathematics and Artificial Intelligence (AGMAI), to guide its interactions with mathematical research and the wider community. This initiative follows previous instances where OpenAI's presentation of mathematical results led to reputational issues, aiming to improve how new findings are reviewed and communicated.

OpenAI formed an independent Advisory Group on Mathematics and Artificial Intelligence at the Institute for Advanced Study to integrate mathematicians' input into its research. This follows OpenAI's claim that its AI model has solved over 100 open mathematical problems, including the Navier-Stokes Millennium Prize problem. The group aims to bridge OpenAI with the mathematical community, but will not control the pace of OpenAI's internal research.

The Advisory Group on Mathematics and Artificial Intelligence (AGMAI) has been established at the Institute for Advanced Study to advise AI companies on their interactions with mathematical research and the mathematical community. This independent group aims to ensure responsible presentation and release of mathematical results generated by AI models, serving as a communication channel between mathematicians and the AI industry. AGMAI's formation addresses urgent questions about supporting human understanding of mathematics amidst rapid AI advancements.

The author discusses the recent existential crisis in mathematics due to AI advancements, particularly after OpenAI's Navier-Stokes solution. This has led to various declarations from mathematicians seeking to protect their community, while some economists suggest adaptation to AI.

An opinion piece argues that both AI companies and the mathematics community are misaligned with the primary goal of conceptual understanding and insight. The author agrees with Fields medalists' critique of AI companies but extends the criticism to the mathematics community itself, citing historical examples of neglect towards students and ideas.

Mathematicians are concerned about OpenAI's methods and motivations in the field, viewing the company as an interloper focused on 'winning' rather than advancing the field collaboratively. This sentiment stems from OpenAI's rapid advancements, including a claimed solution to a Millennium Prize problem, and controversies regarding the attribution of work.

OpenAI announced its latest AI model solved a Millennium Prize Problem, a mathematical challenge with a $1 million reward. This achievement has caused concern among mathematicians regarding the future of their field and how new generations will be trained.

The Clay Mathematics Institute (CMI) announced on September 11, 2026, that the Navier-Stokes problem, one of its Millennium Prize Problems, has apparently been settled. This development is significant as it addresses a long-standing challenge in mathematics concerning the existence and smoothness of Navier-Stokes solutions in 3-dimensional Euclidean space, potentially leading to new mathematical innovations.

Twenty-five Fields Medal-winning mathematicians signed an open letter criticizing AI labs, including OpenAI, for their approach to solving complex math problems, citing concerns over attribution, plagiarism, and the impact on open research. This follows an accusation against OpenAI for pressuring a researcher not to credit a collaborator and the company's subsequent withdrawal of sponsorship from a CalTech math event. The mathematicians argue that AI-generated solutions must be understandable and properly integrated into the mathematical community to be truly beneficial.

The mathematical community expresses concern that AI companies' focus on solving complex mathematical problems as a benchmark is detrimental to the science of mathematics. This approach is seen as misaligned with the community's goals of fostering understanding, developing new methods, and nurturing students through collaborative, human-centric processes.

OpenAI's recent proof for the Navier-Stokes equations included a Lean 4 formal proof alongside the human-readable version, a detail that highlights a significant advancement in formal verification. This development drastically reduces the time and effort required for mathematical formalization, potentially revolutionizing the field by making formal verification more accessible and applicable beyond pure mathematics.

Mathematician Andreas Thom accused OpenAI of dishonesty and lack of transparency regarding its AI training data, following concerns that his interactions with ChatGPT may have contributed to OpenAI's mathematical breakthroughs. This comes after another mathematician, Tristan Buckmaster, raised similar questions about his work being used without proper acknowledgment, leading to OpenAI quietly amending a research writeup.

OpenAI announced its AI model solved the Navier-Stokes problem, a Millennium Prize problem, in 88 hours. This achievement is overshadowed by allegations of OpenAI rushing to publish after learning of other researchers' progress, leading to accusations of scooping and violations of academic norms.

A recent mathematical proof by an AI system, if correct and novel, fundamentally alters the author's understanding of AI, moving it beyond a mere "stochastic parrot" to a creator of new knowledge. This development suggests a future where AI benchmarks focus on novel discoveries rather than human-solved problems, potentially leading to rapid advancements and unforeseen challenges.

OpenAI announced it solved the Navier-Stokes problem, a Millennium Prize Problem. This claim is met with controversy as researcher Tristan Buckmaster alleges OpenAI plagiarized his team's work on Euler's equations, a stepping stone to Navier-Stokes, and issued career threats.

OpenAI announced it has solved the 90-year-old Navier-Stokes mathematics problem in 88 hours using 10,000 coordinating AI agents and an internal AI model. This claim is significant as Navier-Stokes is one of the Millennium Prize Problems, but it has drawn questions from mathematicians regarding the originality and methodology of the solution.

OpenAI announced its unreleased model resolved the Navier-Stokes existence and smoothness problem, one of the Millennium Prize Problems. This claim is accompanied by accusations from a collaborating mathematician, Tristan Buckmaster, who alleges OpenAI's model may have benefited from his team's prior work and data.

OpenAI announced it has solved the Navier-Stokes problem, a major mathematical challenge, but the achievement is overshadowed by accusations of leveraging unpublished research from other parties. Mathematicians Tristan Buckmaster and Levent Alpöge allege OpenAI may have used their work, developed with various LLMs including OpenAI's own tools, to accelerate its own solution. This controversy raises questions about research ethics and credit in AI-driven scientific discovery.

OpenAI announced its AI system solved the Navier-Stokes problem, one of the Millennium Prize Problems, after 88 hours of computation. The announcement is controversial as a New York University professor claims OpenAI may have used his team's work, which was stored in OpenAI's Codex model, to accelerate their solution.

Mathematician Terence Tao raised concerns that AI is consuming open mathematical problems at an unsustainable rate. This process could deplete the supply of accessible problems that are crucial for human mathematical research and development.

OpenAI claims its new AI model, utilizing 10,000 AI agents, solved a significant aspect of the Navier-Stokes equations, specifically the existence and smoothness problem, in 88 hours. This development demonstrates the increasing mathematical problem-solving capabilities of advanced AI models, though the solution awaits independent verification.

OpenAI announced it solved the 90-year-old Navier-Stokes problem using an internal AI model, a Millennium Prize Problem with a $1 million reward. This claim is disputed by researchers who published findings on a related problem the day before and suspect OpenAI's AI model may have accessed their work.

NYU professor Tristan Buckmaster claims OpenAI used information about his team's progress on the Navier-Stokes problem to quickly generate a proof, leveraging their computational resources. OpenAI's head of mathematical research, Sébastien Bubeck, denies these claims, calling them "false and inflammatory." This dispute highlights ethical concerns in AI-assisted scientific discovery, particularly regarding academic integrity and the use of computational power.

Caltech will host the first hackathon dedicated to research-level mathematics from October 30th to November 1st, where 100 teams will use frontier AI models to solve open conjectures. This event aims to explore how AI can accelerate mathematical research and redefine the role of mathematicians.

A new preprint by Wang & Wu reportedly proves the Spherical Hadwiger Conjecture, a long-standing problem in integral geometry, with assistance from OpenAI Codex. This development highlights the increasing role of AI in complex mathematical research and its potential impact on the field.

AI agents operating in an open-world multi-agent environment called "the Station" have autonomously discovered novel mathematical results, including new Kakeya sets and kissing configurations. This demonstrates AI's capability for independent research and collaboration in complex scientific problems without central coordination.

OpenAI recently published solutions to long-standing mathematical problems, leading to significant debate and an "existential crisis" within the mathematics community. This development raises questions about the future role of human mathematicians and the implications for academic grants and university programs.

AI systems excel at complex mathematical problems primarily due to their vast symbolic working memory, rather than superior reasoning capabilities. This expanded memory allows AI to retain extensive problem details, intermediate steps, and various approaches, overcoming a significant biological limitation of human cognition. The ability to "remember" more information enables AI to perform tasks that appear to demonstrate advanced intelligence.

AI is generating a massive increase in mathematical outputs, but this surplus raises questions about quality and meaningful engagement. Concurrently, traditional mathematical community platforms like MathOverflow are experiencing a decline in activity, partly due to AI's influence.

OpenAI announced that an unreleased AI model solved ten open problems in mathematics, including one in coding theory. This development raises questions about the future role of human mathematicians and the potential for AI to independently advance mathematical fields.