Net Good IndexSubmit a correction

Benefit Ledger · provisional · Mathematics

Freitag and Mutchnik, using ChatGPT 5.6, counterexample the stable forking conjecture

On 31 August 2026 James Freitag and Scott Mutchnik posted arXiv:2609.00436, a counterexample to the 1996 Hart–Kim–Pillay stable forking conjecture, found using ChatGPT 5.6.

31 Aug 2026Tier 3 MajorMethodology 0.1

Current score

+0.66

10 base · Major (tier 3 of 5, 10 pts)
× 0.6000 attribution · Critical contribution
× 0.3500 evidence · Technical disclosure or preprint
× 0.4500 realization · Experimentally validated
× 0.7000 durability
Event-level product before credit split: 0.66

A counterexample to a 1996 model-theory conjecture is major (tier 3) if it holds. Humans are the authors; ChatGPT 5.6 found the example (0.60). Evidence is a fresh arXiv preprint (0.35). Realization is a public construction, not community acceptance (0.45).

What happened

The stable forking conjecture asked whether forking independence in simple theories is controlled by stable formulas. Freitag and Mutchnik report a counterexample found with ChatGPT 5.6. The preprint is unformalized and not yet refereed. Model identity is the ChatGPT 5.6 product line, not GPT-6 Astra and not a specified Sol/Terra/Luna SKU.

Model attribution

GPT-5.6
Version unspecified
+0.66

GPT-5.6

Used as ChatGPT 5.6 to find the stable-forking counterexample.

The abstract says the counterexample was found using ChatGPT 5.6; it does not name Sol, Terra, or Luna, so the record stays at the 5.6 generation.

Attribution 0.6000 · Credit share 100% · OpenAI

Claims

  • Freitag and Mutchnik posted a counterexample to the Hart–Kim–Pillay stable forking conjecture, found using ChatGPT 5.6.

    outcome · supported

Sources

primary sources

Revision history

  • 13 Sep 2026 · 0.00 → 0.66

    Imported events under methodology 0.1.