Early science acceleration experiments with GPT-5 — mentioned by vibecodermcswag

Wed, 28 Jan 2026 15:47:29 GMT

Formalized the Sawhney-Sellke stability theorem for Erdős Problem #848 (squarefree products) in Lean 4. ~3800 lines, compiles with 0 errors, no sorries, no axioms. The math was solved by Sawhney & Sellke (Nov 2025): https://arxiv.org/abs/2511.16072 Only ~17 Erdős problems have full solutions formalized in Lean. If valid, this would be ~#18. Done with multiple AI tools (Claude Opus 4.5, GPT-5.2, Gemini 3.0, Aristotle) + human orchestration. Seeking review: Does this actually prove what it claims? The code is verbose and could be cleaner. Would appreciate expert eyes.

Paper and discussion history. A mention is not an endorsement.

Early science acceleration experiments with GPT-5 — HN submission

Sat, 22 Nov 2025 20:39:55 GMT

Early science acceleration experiments with GPT-5

4 points; 1 discussion reply on HN. Paper and source history. A submission is not an endorsement.

Early science acceleration experiments with GPT-5 · HackerPapers discussions

Early science acceleration experiments with GPT-5 — mentioned by vibecodermcswag

Early science acceleration experiments with GPT-5 — HN submission