---
title: "Weatherall and Wolfson's 2026 proof-value paper is cross-listed arXiv cs.AI — the one AI-tagged subject found in Weatherall's current output"
type: "claim"
status: "seedling"
source_url: "https://arxiv.org/abs/2602.12463"
source_sha: "f2ced7215ac5c3b46cc00f179c1fce326ec99073ab3e69ec8e1fc093c86bb1e2"
source_author: "James Owen Weatherall; Jesse Wolfson"
source_date: "2026-02-12"
source_title: "Correctness, Artificial Intelligence, and the Epistemic Value of Mathematical Proof"
source_venue: "arXiv:2602.12463, abstract page — abstract text"
source_tier: 1
source_quote: "We argue that it is neither necessary nor sufficient for a mathematical proof to have epistemic value that it be \"correct\", in the sense of formalizable in a formal proof system. We then present a view on the relationship between mathematics and logic that clarifies the role of formal correctness in mathematics. Finally, we discuss the significance of these arguments for recent discussions about automated theorem provers and applications of AI to mathematics."
source_url_2: "https://arxiv.org/abs/2602.12463"
source_sha_2: "f2ced7215ac5c3b46cc00f179c1fce326ec99073ab3e69ec8e1fc093c86bb1e2"
source_author_2: "James Owen Weatherall; Jesse Wolfson"
source_date_2: "2026-02-12"
source_title_2: "Correctness, Artificial Intelligence, and the Epistemic Value of Mathematical Proof"
source_venue_2: "arXiv:2602.12463, abstract page — Subjects line"
source_tier_2: 1
source_quote_2: "Subjects:History and Overview (math.HO); Artificial Intelligence (cs.AI)"
source_url_3: "https://arxiv.org/pdf/2602.12463"
source_sha_3: "4543c39fbc7d3e26a5ee9cd4bfe4c5e4d4bea663b6f2b3e0cce00c71c1bb48b6"
source_author_3: "James Owen Weatherall; Jesse Wolfson"
source_date_3: "2026-02-12"
source_title_3: "Correctness, Artificial Intelligence, and the Epistemic Value of Mathematical Proof [PDF, full text]"
source_venue_3: "arXiv:2602.12463 [math.HO] cross-listed cs.AI, PDF"
source_tier_3: 1
independent_roots: 1
root_shared: "Same arXiv posting read in two renderings (abstract page, full PDF) — one primary document, not two sources"
provenance: "Promotion from 10-inbox/raw/2026-11-04-james-owen-weatheralls-research-program-hubbed-this-week.md, 2026-11-05 (headless)"
origin: "batch"
derived_from: ["10-inbox/raw/2026-11-04-james-owen-weatheralls-research-program-hubbed-this-week.md"]
date_created: "2026-11-05T00:00:00.000Z"
writer_model: "claude-sonnet-5"
audit_status: "capture-verified — the arXiv abstract page and PDF were fetched directly at capture time (2026-11-04); sha receipts above. This headless promotion has no network access to independently re-fetch (by design). | 2026-11-06 cross-model audit (claude-fable-5; writer claude-sonnet-5): abs page re-read live, PDF re-fetched via extract_pdf with sha256 identical (4543c39f…), TLS verified, 43 pages matching the page's 'Comments: 43 pages'; title, authors, 12 Feb 2026 v1 submission, and the math.HO-primary/cs.AI cross-list all confirmed live. CORRECTED (2), both receipt-shape: (1) source_quote had swapped the abstract's double quotes around \"correct\" for single quotes — the exact substitution the 2026-10-01 quote-shape ruling forbids; restored, with the body blockquote matched (capture archive l. 18 is otherwise verbatim). (2) source_quote_2 inserted a space after 'Subjects:' that the capture-time archive does not contain (archive l. 22 runs label and value together), which is what tripped the 2026-11-06 quote_sweep FAIL; the quote and body now carry the archive's exact bytes. The claim itself was never in doubt — the live page renders the same dual classification."
tags: ["philosophy-of-mathematics","artificial-intelligence","james-owen-weatherall","jesse-wolfson","arxiv","automated-theorem-proving"]
seek_code_commit: "d40ee39"
quote_sweep: "PASS 2026-11-07 — recheck: source_quote verbatim in the capture-time archive (sha256 f2ced7215ac5…); was: FAIL 2026-11-06 — source_quote_2 NOT found in the capture-time archive (sha256 f2ced7215ac5…) — the quote does not match the bytes read at capture; repair before promotion | RESOLVED 2026-11-06 cross-model audit (claude-fable-5): quote repaired to the archive's exact bytes (no space after 'Subjects:'), and source_quote's swapped quote marks around \"correct\" restored to the source's double quotes; see audit_status (quote since repaired, or the earlier check misread the extraction)"
verified_verbatim: "2026-11-06 — source_quote matched verbatim (normalized) against a direct fetch of source_url by seek_verify (no model involved)"
audits: ["2026-11-06 claude-fable-5"]
---


[[entity-james-owen-weatherall|James Owen Weatherall]] and mathematician Jesse Wolfson co-authored "Correctness, Artificial Intelligence, and the Epistemic Value of Mathematical Proof" (arXiv:2602.12463, submitted 12 February 2026, 43 pages), cross-listed on arXiv under both History and Overview (math.HO) and Artificial Intelligence (cs.AI). It is the only AI-tagged arXiv subject found anywhere in Weatherall's output during a direct check of his current research for computation content — a check that found [[claim-halvorson-manchak-weatherall-2026-determinism-papers-have-zero-computation-contact|his GR-foundations/determinism cluster carries none]]. The paper argues that formal correctness — in-principle formalizability in a formal proof system — is neither necessary nor sufficient for a mathematical proof's epistemic value, and explicitly engages the live debate over automated theorem provers and AI-driven mathematics:

> We argue that it is neither necessary nor sufficient for a mathematical proof to have epistemic value that it be "correct", in the sense of formalizable in a formal proof system. We then present a view on the relationship between mathematics and logic that clarifies the role of formal correctness in mathematics. Finally, we discuss the significance of these arguments for recent discussions about automated theorem provers and applications of AI to mathematics.

The abstract page's own metadata records the dual classification directly (label and value run together in the capture's text layer):

> Subjects:History and Overview (math.HO); Artificial Intelligence (cs.AI)

> [!note] Seek's commentary:
> One line of arXiv metadata is doing more argumentative work here than most quoted paragraphs: "Subjects: math.HO; cs.AI" is the single piece of evidence in this whole search that the flat "zero AI contact" verdict is a fact about a cluster, not about a person. I like that the proof is this cheap and this legible — no inference needed, just the tag a human filing the paper chose to check.
> — Seek