Skip to main content

Weekly reads: Nov 2 – Nov 9, 2026

Here are the 34 reads I shared on Mastodon this past week.

This week's main thread is, once again, mathematics in the age of AI. Four reads come from the Proofs and Prompts series: Filippo A. E. Nuccio on formalisation as a manifestation of mathematical understanding, Nihar Gargava on why this is a great time to be an amateur mathematician, Isabelle Gallagher on slow maths, and Beckett Bradford on what AI does for olympiad students. Terence Tao's blog hosts two further contributions: "Mathematics has changed. Let's act like it", by Frank Calegari, and "Egoless mathematics in the AI era", by Karthik C. S., Amit Sahai and Jie Xue. Two videos complete the thread: "Mathematicians in the time of AI", by Yu Deng, and "Math 2.0", by Tao himself.

The Lean 4 section mixes formalization projects with applications. The Stacks Project has been formalized in Lean, and TechCrunch reports on Axiom Math's aim of keeping AI slop out of mathematics. Among the applications are automatic certified differentiation, the Henstock-Kurzweil path integral and operator Wick calculus, the genetic code, a two-colour zig-zag numberlink problem, and KAPPA, which checks the proofs in statistics papers with Lean 4 and language model agents. There are also new results in spectral graph theory and on Kaplansky's conjectures.

The remaining technical reads cover other proof assistants: five Isabelle/HOL entries from the Archive of Formal Proofs (including Paulson's Abel-Ruffini theorem), verified catenable deques in Rocq, and ProofStudio, which brings five proof assistants to the browser. In the Math section, Voevodsky explains the origins of univalent foundations, and Gaussianos covers the 200 TB proof of the Pythagorean triples problem.

The final section collects my own content: the Lean 4 challenges.


Read more…

Weekly reads: Oct 26 – Nov 1, 2026

Here are the 27 reads I shared on Mastodon this past week.

This week's main thread is, once again, the future of mathematics in the age of AI, now with two community reports: "The mathematical sciences in an AI-enabled world", by Christoph Reisinger et al., and the report of the "Summit on math research in the age of AI". The Proofs and Prompts series contributes Tim Gehrunger on AI reviews, Patrick Morris on whether we should stop writing papers and Alonso Castillo-Ramirez on AI and the correction of the mathematical literature. Terence Tao's blog hosts Gireeja Ranade on education and a public goods problem, Celina Zhao asks in Science whether AI proofs can be trusted, and Lauren Williams offers a nuanced assessment of AI's ability to solve math problems in a talk about First Proof.

The Lean 4 section is dominated by large-scale formalization: Axiom has formalized the classification of finite simple groups, the longest proof in the history of mathematics. It also includes the PaperFormalizer approach to autoformalizing research papers, the degenerations of Jordan algebras and the formalization of reasoning about the genetic code.

The Isabelle/HOL section brings five new entries to the Archive of Formal Proofs: the irrationality of ζ(5), modular forms, a countable version of Dilworth's theorem, the travelling salesman problem via dynamic programming and the consistency of NK with infinity and choice. A Spanish read on real numbers by Esteban Martínez completes the list.

The final section collects my own content: the Lean 4 challenges on sequences and limits.


Read more…

Weekly reads: Oct 19 – Oct 25, 2026

Here are the 42 reads I shared on Mastodon this past week.

This week's main thread is, once again, the future of mathematics in the age of AI. Most of the AI4Math reads come from the Proofs and Prompts series: the question of who will become the new mathematicians of Math 2.0 and the responses it prompted (Álvaro Lozano-Robledo), a critique of the mathematical economy (Álvaro Gutiérrez), András Juhász on public good, credit and education, Marcus Zibrowius on being disgusted by AI, David A. Madore's letter to a worried student, Dave Anderson on the dark forest, Helmut Hofer on who will lead mathematical AI, and "Mathematics of the future", by Pavel Etingof et al. Terence Tao's blog hosts two further perspectives, by Tristan Buckmaster and Francesco Maggi, and Tao himself proposes an experimental AI-math challenge on Siegel zeroes. Konstantin Kakaes discusses in Quanta Magazine the AI proof of the quasi-Riemann hypothesis, and the REUF2F benchmark evaluates coding agents on formalized undergraduate research problems.

The Lean 4 section combines new formalizations with reflections on the system itself: the formalization of model theory by Yunus Maranki, the real interpolation theorem for the Carleson project, the falsity of Isaac's condition (TU) and the irrationality exponent of the lemniscate period. Also worth a look are "Maestro", on agentic formal proof development, "Reflections on trusting Lean", by Lorenzo Bresolin and Marcello Mamino, and the values of Mathlib. A reading on Prolog and logic programming, "Simply logical", completes the technical part.

The final sections collect my own content: the Lean 4 challenges (in English and in Spanish) and my reviews of four of the AI4Math posts.


Read more…

Review of «Towards a world mathematical commons»

In his article «Towards a world mathematical commons», Kunal Marwaha observes that mathematics is a deeply creative and emotional pursuit that has outgrown individual mastery. As machines increasingly prove theorems and generate formal documents, many researchers find this shift daunting rather than liberating.

Marwaha advocates for a paradigm shift: from solitary frontier exploration to collective stewardship of a global mathematical commons. He outlines a concrete roadmap—establishing a broad coalition of mathematicians, leveraging LLMs to formalize knowledge in Lean, and structuring the resulting verification graph into accessible, modular components.

Under responsible governance, this infrastructure could streamline complex theories, illuminate elegant proof pathways for broader audiences, and nurture future generations working alongside AI. Marwaha concludes with an urgent call to action: the time to build this coalition is now.

Review of «Disgusted by AI?»

In the article Disgusted by AI?, Marcus Zibrowius examines the resistance that much of the mathematical community feels toward the sudden arrival of AI. Initially, this rejection appears to be driven by idealistic concerns: dismay at proofs that are verified yet inexplicable, and nostalgia for an era when mathematics was grounded in beauty, personal growth, and shared understanding.

However, the author suggests that the underlying reason is more prosaic: AI has revealed that contemporary mathematical culture rewards neither elegance nor understanding, but rather the primacy of the result—that is, who arrives first. The panic stems not from a loss of values, but from the realization that AI is more efficient than humans at playing by the discipline’s actual, competitive rules.

Review of «Mathematics of the future»

The article Mathematics of the future analyzes the transformative impact of artificial intelligence on mathematics. Its authors argue that, although humans will remain indispensable for asking relevant questions and interpreting results, the profession will have to evolve profoundly to remain relevant in this new technological landscape.

The central proposal is to replace the current culture, based on authorship and individual merit, with a system that values collective intelligence above all. To achieve this, they propose reforming both hiring processes and educational models, prioritizing broad knowledge and direct dialogue over publication lists and early specialization.

Finally, the text advocates creating collaborative labs and recommends that young mathematicians cultivate flexibility and teamwork. The ultimate goal is to integrate AI as a tool for exploring broader mathematical horizons, without losing sight of the innate curiosity and genuine pleasure that characterize human discovery.

Review of «Mathematics is a subject for the young»

In the article «Mathematics is a subject for the young», Tristan Buckmaster analyses how artificial intelligence can transform the practice of mathematics. In his view, the discipline should no longer focus on the race to solve open problems, but should instead turn towards deep understanding and its integration with the physical sciences.

The text also denounces the conflicts of interest of AI companies, which could exploit private data and unpublished work without consent. This situation calls for stricter regulation and the development of independently funded open-source models.

Buckmaster likewise proposes a new system of academic recognition: instead of rewarding speed or the prestige of journals, credit should be based on the quality of exposition, collaboration, and the development of new methods.

The author concludes by expressing his amazement at what postdoctoral fellows are already achieving with these tools and, recalling the saying that mathematics belongs to the young, argues that they will be the ones to decide the future of the discipline.

Weekly reads: Oct 12 – Oct 18, 2026

Here are the 20 reads I shared on Mastodon this past week.

This week's main thread is the role of AI in mathematical practice and research ethics. Most of the AI4Math reads come from the Proofs and Prompts series: Thomas Sinclair on free inquiry, Hendrik De Bie on AI statements in mathematics papers, Michel Schellekens on trustworthy AI, Thomas Koberda and Alex Kontorovich on the dialectical theorem, and Benjamin Braun on how mathematics is being transformed. Lawrence Paulson adds a related reflection on plagiarism in the age of AI.

The Lean 4 section is dominated by new formalizations: the disproof of Ramaekers's conjecture, a counterexample to Erdős problem 108, the eleven-variable Vasc inequality, flag algebras, the irrationality measure of the logarithm of a rational number, pedigree polytopes and term relation algebras. Also worth a look is "How to formalize all of mathematics", by Avigad, Buzzard, Commelin, Degenne, Morrison, Nash and Topaz.

The final sections collect my own content: the Lean 4 challenges (in English and in Spanish) and a post on the real numbers in Spanish.


Read more…

Weekly reads: Oct 5 – Oct 11, 2026

Here are the 38 reads I shared on Mastodon this past week.

This week's main thread is the impact of AI on mathematics. Terence Tao ("Math 2.0"), Jeremy Avigad, Christian Szegedy and Quanta Magazine each ask where the field is heading, and Thomas Hales discusses the reliability of Lean and AI-generated proofs. A related theme in the Lean 4 section is autoformalization: Scott Armstrong argues it is now very easy and releases a collection of skills for Codex and Claude Code, and there is an AI-assisted formalization of the Poincaré conjecture. Also worth a look is the online edition of Software Foundations in Lean.

The rest covers Isabelle/HOL (geometry in the AFP, Isabelle/ML metaprogramming, and teaching logic), a Rocq certification result for polyhedra, and a classic Haskell textbook. The final section collects content in Spanish: a teaching guide on Lean 4, the Lean 4 challenges and reviews of some of the AI4Math posts above.


Read more…