What was already known
Pettie and Tardos refuted the stronger n·polylog(n) conjecture using acyclic patterns, but the weak n^(1+o(1)) version remained open.
AI involvement
Gishboliner and Li credit ChatGPT-6 Astra with finding the proof under substantial guidance. They rewrote the argument and explain its entropy and ordered-tree steps.
The authors expressly credit ChatGPT-6 Astra with finding the proof under substantial human guidance, followed by their rewrite.
Importance
Resolves the weak acyclic-pattern question with a quantitative bound.
Limitations
The stronger polylogarithmic claim remains false. The authors rewrote the model-found proof, but do not describe a completed technical check of the final argument.
Technical verification
The manuscript includes a proof, but author responsibility and rewriting do not specify a completed technical check or method.
Human understanding
The authors rewrote and present the proof’s central entropy and ordered-tree argument; no unexplained core step is reported.
Correction and publication history
Resolves the weak acyclic-pattern question with a quantitative bound.
Read this published revision