AI-Generated Proofs Deserve Public Posting, Argues Researcher, Readers Can Just Ask AI to Rewrite Them

aran_nayebi · x · 2026-09-29

Sharing researcher arannayebi's endorsement of PI010101's take: if someone uses AI in any way to solve a problem at least one person genuinely wanted solved, publishing the AI-generated paper under their own name on arXiv or GitHub is perfectly fine.

He argues it's a misconception that mathematicians must read AI proofs verbatim and then complain about "AI slop" — instead, they should simply ask AI to rewrite the proof in their preferred language, style, and mathematical tools, and after a few prompts get a readable version.

Having a valid proof publicly available is already valuable in itself: it gives AI something concrete to build on.

Original post →

More from AGI Musings

AGI Musings channel →