Osserman, Matsaev and more: Goel’s AI-written claims
Awaiting review
The takeaway
In three days Dave Goel deposited sixteen Hexagon works (most results also in an expanded “and other miscellanea” version) stating that they were ‘done overwhelmingly by AI models’ (Claude and ChatGPT) and only partly checked by a human. The headline claims: the Osserman conjecture holds in every dimension, and Matsaev’s conjecture fails for every p ≠ 2 already for a 2×2 matrix. None is refereed.
Status
- Claim
Awaiting review
- Our reporting
- Low confidence
- Verification
- Unverified claim
- Importance
- 3 of 5
- Last verified
- 19 November 2026
Your AI and this story
- GPT-6.2 Astra170 days after its cutoff
- Claude Opus 5.5140 days after its cutoff
- Gemini 4 Hydrogen231 days after its cutoff
- Grok 4.7170 days after its cutoff
None of these four assistants can know about it. The closest, Claude Opus 5.5, stops 140 days before it.
Key facts
- Osserman (2611.00544, abstract): ‘Chi and Nikolayevsky proved it in every dimension except one case in dimension sixteen… We prove this conjecture. Hence the Osserman conjecture is true in every dimension, and so is the conformal Osserman conjecture’; expanded version on Zenodo (doi 10.5281/zenodo.23603278)
- Matsaev (2611.00547): Drury (2011) had disproved it for p = 4 and 4/3 with a computer-assisted example; the deposit claims failure for every p ≠ 2 ‘without numerical computation’, and that the inequality holds for quadratics whose constant and leading coefficients have equal modulus
- Wills (2611.00540): Wills proved n ≤ 3, Henze (2014) showed failure for n ≥ 7; the deposit settles n = 4 (true), 5 and 6 (false)
- Also: Lee’s 2026 conjecture on diameters of order and chain polytopes fails in both directions (2611.00525); nowhere-zero flow reconfiguration for every abelian group of order ≥ 3024 (2611.00522); flow reconfiguration and Cereceda’s conjecture for bipartite 2-degenerate graphs, 181 pp. (2611.00520); companion ‘and other miscellanea’ versions (2611.00541, 2611.00546, 2611.00548, 2611.00513, 2611.00527); the Esperet–Hendrey–Lagoutte–Marseloo–Norin–Steiner lifting question for k = 2,…,6 (2611.00521) and the poset-diameter construction B_k as a standalone paper (2611.00523), both 16 Nov
- Provenance (verbatim, Osserman deposit): ‘This work was done overwhelmingly by AI models and supervised by the author. The models were Anthropic’s Claude and OpenAI’s ChatGPT. The author takes full responsibility for the paper.’ Human understanding: ‘Some parts of this contribution, but not all, have been checked and understood by a human.’
- Earlier Goel claims in the dataset: Smale’s mean value constant equals 4 with a Lean proof (entry 2026-11-11), Sheil-Small’s covering problem and Cereceda’s conjecture (listings of 7 and 10 Nov)
- 19 Nov deposit: ‘The Černý (1969) conjecture holds for quasi-Eulerian automata of degree one’ (Hexagon 2611.00635): every synchronizing quasi-Eulerian automaton of degree one has a reset word of length at most (n−1)², improving Berlinkov’s 2^d(n−d+1)(n−1) for this family; provenance: ‘done overwhelmingly by AI models, steered by the author in a continuous critical dialogue, alternating between Anthropic’s Claude and OpenAI’s ChatGPT’; human understanding ‘some’
What happened
An independent author posted a series of deposits, each in a short and an expanded “and other miscellanea” version, that claim to settle named problems across geometry, operator theory and combinatorics. All are described as mostly AI-generated with partial human checking.
Why it matters
The Osserman conjecture had been reduced to a single algebraic case in dimension 16, and Matsaev’s conjecture is a long-standing question in operator theory. Claims of this kind in this volume from one contributor are hard for experts to check, and none has been examined publicly yet.
What is disputed or not yet verified
| Verification | Unrefereed Hexagon deposits; no Lean for these results |
|---|
Sources
10 sources from 1 site. Numbers match the chips in the text.
10 sources: 10 primary
Primary
- Hexagon 2611.00544: The Osserman (1990) conjecture is fully resolvedhexagonmath.org, paper
- Hexagon 2611.00547: The Matsaev (1971) conjecture is false for every p ≠ 2hexagonmath.org, paper
- Hexagon 2611.00540: Wills’ (1981) conjecture holds iff the dimension is at most 4hexagonmath.org, paper
- Hexagon 2611.00512: The Hibi–Li (2016) conjecture is fully resolvedhexagonmath.org, paper
- Hexagon 2611.00526: the Nagata–Higman index is at most n² − 2hexagonmath.org, paper
- Hexagon 2611.00635: The Černý (1969) conjecture holds for quasi-Eulerian automata of degree onehexagonmath.org, paper
- Hexagon 2611.00546: The Osserman (1990) conjecture is fully resolved, and other miscellaneahexagonmath.org, paper
- Hexagon 2611.00548: The Matsaev (1971) conjecture is false for every p ≠ 2, and other miscellaneahexagonmath.org, paper
- Hexagon 2611.00513: The Hibi–Li (2016) conjecture is fully resolved, and other miscellaneahexagonmath.org, paper
- Hexagon 2611.00527: Nagata–Higman index at most n² − 2, and other miscellaneahexagonmath.org, paper
Changes
- Added the Černý deposit 2611.00635 and full links for the four ‘miscellanea’ records, which our automated search did not count as cited (science agent)
- Full Hexagon ids for the miscellanea versions; added 2611.00521 and 2611.00523
- Filed (science agent; provenance records and abstracts read through the Hexagon API)