AI & Maths watch GitHub

Retrospective research digest

Updates on artificial intelligence for mathematical research, proof discovery and formal verification, with links to papers, code and tools.

At a glance1 entry · 2 linked sources Partial coverage

One substantive essay was established as published on September 12, 2026. No new mathematical result, proof artifact or research tool met the date and evidence requirements in the accessible material. Archival coverage remains incomplete.

Research updatesSelect an entry to read more

Manuel Rivera distinguishes AI's effects on mathematical understanding and the research profession Manuel Rivera Published: 2 sources

Rivera separates mathematics as a domain of ideas, the human pursuit of structural understanding, and the institutional profession. He identifies communication, writing, counterexample searches and literature navigation as potentially useful AI applications, while arguing that commercial incentives, unreliable output and misuse may reshape the profession more profoundly than the pursuit of understanding itself.

Research relevance:

Relevant to researchers deciding which activities to delegate to AI and how to preserve mathematical judgment, graduate training and incentives for understanding. It provides a substantive perspective on research practice rather than announcing a theorem or tool.

Claim status:
Signed opinion essay read on the original publishing site. Its institutional predictions and judgments are the author's arguments, not experimentally established findings. No proof or formalization is claimed.
Limitations:
The site displays publication on September 12, 2026, and separately records receipt on August 23, 2026. No publication time or time zone is supplied. The essay offers no reproducible workflow, benchmark or technical validation of the suggested AI uses. Reader comments were not treated as evidence.

Sources & further reading

Link to this entry
Coverage and limitations
  • Public web searches reconstructed September 12, 2026, the requested Paris civil day. Inclusion relied on dates displayed by original sources, not search-index or crawl dates. Publication times and source time zones were generally unavailable.
  • Read Manuel Rivera's dated essay on Proofs and Prompts and followed its link to his personal website. The canonical article URL returned an error; the article was readable through its comment-page URL.
  • Read Epoch AI's September 12 newsletter and its linked FrontierMath Erdős announcement. The underlying announcement is dated September 1, 2026. Its benchmark and solution claims were excluded as older material recapitulated on the target day; no genuine September 12 update was established.
  • Read Oliver Knill's Harvard-hosted note, explicitly posted September 12, about Javier Gomez-Serrano's September 11 talk. It primarily comments on previously supplied fluid-equation announcements and adds no accessible technical proof assessment substantial enough for a separate entry.
  • Consulted Terence Tao's What's New and the Xena blog. Buzzard's Imperial College homepage explicitly identifies https://xenaproject.wordpress.com/ as his blog. The previously supplied September 11 declaration and September 7–10 fluid-equation coverage were excluded from this edition.
  • Consulted official websites for Lean, Project Numina, Harmonic, Axiom and Epoch AI. Project Numina's homepage yielded no readable text. These visits did not establish a qualifying September 12 release. Official sites were used to check organizational provenance; the supplied social handles were not all independently authenticated.
  • Date-specific searches covered the supplied researcher and organizational social-account leads. No exhaustive X archive was accessible. Search results for X and third-party mirrors were leads or indexed snippets, not authenticated readings of complete historical timelines. Mathstodon and Bluesky timelines were not directly consulted.
  • A search snippet for OpenProblemsInNLA described resolutions recorded September 12 and later Lean checks. The opened GitHub page did not expose the relevant file contents, and original publication dates were not established. These claims were excluded; later verification claims were not attributed to the target day.
  • No source code was executed, no Lean proof was checked, and no mathematical proof was independently verified.

Requested coverage window: 2026-09-12T00:00:00+02:00 — 2026-09-13T00:00:00+02:00.

Public web research; coverage is not exhaustive. A source link is not a certification of a claim.

Archives 32
Older editions (2)