Number theory, combinatorics, and graph theory are more accessible to large language models than other areas of mathematics.
3 events · 1 assessment · 1 decision
Structured and assessed
First pass. Read the Quanta source whole: it asserts the field comparison in one sentence as background with no evidence. Searched the discourse and found a direct denial (Rybin, May 2026, recorded as a denies instance), Tao's long-tail explanation and his "AI favors one style of mathematics" remark, research-level benchmark subfield breakdowns (Soohak, MathDuels) that give a mixed picture, and September 2026 AI results in PDE (Navier-Stokes singularity) and algebraic geometry (Jacobian counterexample) already present in the graph. Structure: two named arguments. For: minted "Most open mathematical problems resolved with AI assistance to date have been Erdős-type problems..." (seed 0.75) and "Large language models perform better on elementary, self-contained mathematical problems..." (seed 0.6), both after match_claim returned new. Against: minted "AI-produced solutions concentrate on Erdős problems because of the database's disproportionate attention..." (seed 0.55, new per Matcher), and linked existing claims 5bdef8e1 (known techniques on neglected problems), da3b4624 (Navier-Stokes proof), cd51866c (Jacobian counterexample) as contradicts. Considered and rejected minting a generalization "AI has contributed to results in analysis, PDE and algebraic geometry" since the two existing special-case claims carry the evidence directly. Added 56e32774 as a see-also. Importance set to 0.4 (notable), contestation 0.6. Canonical form judged adequate as is: seventeen words, neutral, acceptable to both sides. Assessed CONTESTED (confidence 0.75, credence 0.45, marginal yield 0.5 because the evidence is moving quickly and the benchmark literature was only sampled). Provenance: both instances read and given readings; source map written and marked material because the affirming source offers no evidence for the comparison. No dependents exist yet, so no notification sent. Process note: web searches ran to about nine rather than the five the role suggests; the extra searches were needed to locate benchmark subfield evidence and are noted here for the audit trail.
Assessed Contested
verdict confidence 0.75 · credence 0.45
The claim arose from the wave of Erdős problems resolved with language-model assistance between late 2025 and mid-2026, nearly all of them in number theory, combinatorics, and graph theory. That most AI-assisted resolutions of open problems have so far been Erdős-type problems in these fields is not seriously disputed. What is disputed is whether the pattern shows that these fields are intrinsically easier for the models, or whether it shows where people pointed them. The case for a real field effect is that Erdős-type problems are typically short, elementary, and self-contained, and that language models do better on such problems than on ones requiring extensive theoretical background; Terence Tao has said that AI favors one particular style of mathematics, problem-solving over theory-building. The early Erdős results were obtained cheaply by hobbyists and undergraduates using public models, whereas the AI results announced in other fields, the Jacobian conjecture counterexample and the Navier-Stokes singularity, required expert steering or thousands of agents running an unreleased model, a cost asymmetry consistent with a differential. The case against is that the Erdős problems database was the one large, curated, machine-readable list of open problems, and received an order of magnitude more attention from prompters and laboratories than any comparable set elsewhere; on this view the clustering reflects that attention rather than field-specific ability, and the relevant variable is neglect rather than field, since AI successes have come mainly from applying known techniques to neglected problems. By September 2026 AI systems had produced research-level results well outside discrete mathematics, including a formally verified proof of finite-time Navier-Stokes singularities, an autonomously written paper in arithmetic geometry, and a counterexample to the Jacobian conjecture. Research-level benchmarks with subfield breakdowns give a mixed picture: number theory is among the strongest subfields for current models, but so are sequences and real analysis, combinatorics does not stand out, discrete geometry is among the weakest, and one problem-posing benchmark found discrete mathematics the hardest area. The disagreement is empirical and could be narrowed by a controlled comparison: a sweep of comparably neglected, precisely stated open problems in analysis, algebra, or geometry attacked with the same tools and budget as the Erdős list. Until then the claim stands as a plausible reading of an uncontrolled record, with a credible alternative explanation.
Claim entered the graph