± HEALTH Colorectal Cancer (CRC) is a common cause of cancer death and prognostic factors are used to determine manage…+ HEALTH This multicenter study developed and validated an interpretable machine learning model integrating granular n…+ HEALTH Multiple stakeholders need to locate results of registered clinical trials but frequently struggle to find th…+ HEALTH Accurate preoperative glioma grading and molecular subtyping are important for treatment. The vascular microe…+ HEALTH Gastrointestinal endoscopy generates extensive high-resolution video data, posing significant challenges for…+ HEALTH Non-invasive follicular thyroid neoplasm with papillary-like nuclear features (NIFTP) and invasive encapsulat…+ POLICY Artificial intelligence (AI) has the potential to strengthen real-world evidence (RWE) for regulatory decisio…± HEALTH Echocardiography remains the cornerstone of cardiovascular imaging. However, traditional workflows including…
TruaceTracing the truth around AIFriday, August 28, 2026
Subscribe
Science·P Space·Evidence-backed problem·Published 2026-08-28

Surprising AI breakthroughs raise soul-searching questions for mathematicians | Letter

Abstract: I share Kasra Rafi and Bruce Schneier’s impression that recent mathematical breakthroughs by AI consist in clever recombination of existing ideas, not development of truly novel theory (No, AI doesn’t mean the end of mathematics – at least not yet, 25 August). The question is: what happens to mathematics if this changes? Like many mathematicians, I have done much soul-searching in recent weeks, especially since a key problem in my own field of group theory (the existence of non-sofic groups) was solved this mont…

TRV-2026-0911JournalismPermanent record — cite & verify
Surprising AI breakthroughs raise soul-searching questions for mathematicians | Letter
The quick read

In a Guardian letter dated 27 August 2026, a Cambridge mathematician reports that OpenAI's Astra model solved the existence of non-sofic groups, a key problem in group theory, using a slight twist on theorems by Gabor Kun and Andreas Thom.

The breakthrough matters because it shifts debate from whether AI can do mathematics to why human mathematics matters, raising uncertainty about whether universities will continue to support human researchers if AI can generate theorems more cheaply, and whether future systems will move beyond recombination to truly novel theory.

Main points
  • Letter writer notes Astra's proof builds on existing work by Gabor Kun and Andreas Thom.
  • Author cites Bill Thurston's 1994 essay On Proof and Progress in Mathematics to argue mathematics aims at human understanding, not just answers.
  • Author warns cash-strapped universities may view human mathematical researchers as superfluous if AI produces papers faster and cheaper.
Problem

If AI can produce mathematical research faster and cheaper, university administrators may treat human mathematics researchers as superfluous.

The rundown

The letter is written by Dr Henry Bradford, Fellow of mathematics at Christ's College, University of Cambridge, responding to recent coverage of AI in mathematics.

The author describes soul-searching among mathematicians after Astra's result and frames the future as a societal choice about what aspects of human intellect to value, not just a technical question of AI capability.

Sources

Reader signal

How should this claim be treated?

The debate