OpenAI changed Navier-Stokes press release and Lean4 code on GH

The reported situation around OpenAI's Navier-Stokes press release and Lean4 GitHub modifications reveals critical tensions between AI-assisted mathematical claims and formal verification standards. Here's the definitive breakdown:

Core Agreement: Why This Matters

1. ·Navier-Stokes is a Millennium Prize problem – Any claim requires extraordinary scrutiny. The mathematical community rightly demands formal verification like Lean4 provides, where correctness is binary (code either type-checks or doesn't).
2. ·Press release/code mismatches undermine credibility – If OpenAI adjusted Lean4 implementations or announcement language post-release, this suggests the initial claims didn’t survive rigorous verification. Such discrepancies aren’t minor—they contradict the entire purpose of using proof assistants.
3. ·AI faces a 'proof gap' in mathematics – Large language models can generate plausible-looking proofs, but Lean4 serves as an unforgiving filter. Post-release modifications likely indicate either incomplete/informal 'heuristic' solutions or outright errors requiring correction.

Key Insights from Both Perspectives

• ·The tension between marketing narratives and technical reality: Bold claims about solving Navier-Stokes (a historic AGI benchmark) may outpace what formal verification actually supports. Silent GitHub updates or press release edits suggest overreach in initial framing.
• ·The distinction between 'solving' and 'formalizing': AI may assist with lemma suggestions or structure imitation, but mechanical verification remains the gold standard. Modifications imply the original work was either trivial (e.g., formalizing known properties) or non-verifiable.
• ·Strategic implications: This incident underscores that AI's role in mathematics isn’t about autonomous 'solutions' but iterative collaboration—LLMs propose, verifiers reject, humans refine. The real progress lies in closing this loop.

Critical Takeaway

In mathematics, 'almost correct' is failure. For AI to meaningfully contribute, Lean4 code must precede press releases—not the reverse. The observed edits reveal a need for stricter alignment between marketing and the immutable rigor of formal proof systems.