VibeMathedMath problems solved with AI

Kahn's flow conjecture

Chvátal conjectured in 1974 that among the intersecting subfamilies of any downset (a family closed under taking subsets), one of maximum size can be taken to be a star: all sets containing some fixed element. Kahn's flow conjecture is a strengthening, phrased as the existence of a flow between the family and a star, which implies Chvátal's statement. Is Kahn's flow conjecture true?

Result
Proved
Status
Resolved
AI contribution
AI co-developed
Method
Argument
Field
Extremal set theory
Posed by
Jeff Kahn, as a strengthening of Vašek Chvátal's 1974 conjecture on intersecting subfamilies of a downset
Year posed
1974
Years open
52y
Solved
2026-09-20
Model
GPT-6 Astra
Vendor
OpenAI
Collaborators
Peter Keevash
Verification
Unreviewed
Publication
Preprint
Significance
40 / 100
Disclosed cost
Wikipedia
No dedicated article

What the AI did

The paper's statement on AI use: the proof was found by GPT-6 Astra, following an approach suggested by the author. The author supplied the route and the surrounding write-up; the model produced the argument along it.

Verification

Checked here on 22 September 2026 against arXiv:2609.23595: the abstract states that Kahn's flow conjecture is proved and describes it as a strong form of Chvátal's conjecture, and the reference list confirms the attribution chain (Chvátal 1974, Hypergraph Seminar, LNM 411; Friedgut, Kahn, Kalai and Keller, JCTA 156 (2018), on Chvátal's conjecture and correlation inequalities). The mathematics was not checked here. Two days old, no referee. A separate September 2026 preprint (arXiv:2609.19123, Chang, Liu and Liu) proves Chvátal's conjecture itself by a correlation inequality and is cited here; the two are independent.

Sources

Changelog1 change

Discussion