What Proof Supports OpenAI’s AI Claim About Exponential Time?

🔍 Read the full analysis: What Proof Supports OpenAI’s AI Claim About Exponential Time? on ThorstenMeyerAI.com

Buying for a business?Offer from Amazon

Get business pricing on monitors, keyboards and dev gear

• Business-only prices and quantity discounts
• Tax-exempt purchasing
• Multiple users, one account, clear invoices

Create a free Business account →Free to register

As an affiliate, we earn on qualifying purchases.Hide

OpenAI said on October 26 that an unreleased AI model produced a proof refuting the Exponential Time Hypothesis (ETH), a long-standing conjecture about the difficulty of solving certain logic problems. The result is available as a preprint and a Lean-checked formalization, but the mathematical claim has not yet received independent expert sign-off, and the formalization covers a weaker runtime bound than the paper’s sharpest result.

The preprint claims a deterministic algorithm can exactly count satisfying assignments for any fixed-width k-SAT formula in subexponential time while using polynomial memory. For 3-SAT, that would contradict ETH, which says no algorithm can solve the problem in subexponential time. The paper also says its result refutes the stronger Strong Exponential Time Hypothesis and three related hypotheses.

OpenAI says its model found the disproof on October 16, with the Lean formalization produced the following day. The preprint is dated October 17 and was revised October 25. The source report describes the public formalization as roughly 3,977 files and 710,000 lines; those figures are the reporter’s count, not an independently stated OpenAI measurement. The formal statement itself is much shorter than the codebase.

Lean is software that checks whether a formal proof follows from its stated definitions and rules. That check supports the formalized statement, but it does not by itself establish that the statement matches every strongest claim in the preprint. According to the source report’s examination of the scope notes, the formalized bound is weaker than the paper’s sharpest runtime bound. The group reviewing the work is led by Russell Impagliazzo, Ramamohan Paturi and Ryan Williams; Impagliazzo and Paturi helped formulate ETH.

A Challenge to SAT Hardness Assumptions

If independent experts validate the result, it would change a widely used assumption in theoretical computer science: ETH underpins proofs that many problems cannot be solved substantially faster than current algorithms allow. Researchers use such hypotheses to explain why certain computational tasks are believed to require exponential time. A refutation could require revisiting results whose conclusions depend on ETH, although it would not automatically provide practical algorithms for those tasks.

The claimed algorithm is subexponential, not polynomial-time. The source material gives its approximate running time as 2 raised to n divided by the square root of log n. That remains far slower than polynomial time, and OpenAI says practical improvements would require significant further research. The result does not solve P versus NP, which remains an open problem. Its immediate importance is therefore a potential shift in theoretical assumptions and an unusual test of how AI-generated mathematics can be checked.

The process matters as much as the headline claim. A formal proof checker can verify that a proof encoded in its language follows specified rules; it cannot replace expert scrutiny of the model, the formalization’s scope or the relationship between the code and the paper. Here, public code and a review led by researchers associated with ETH offer ways to scrutinize the claim, while the gap between the paper’s strongest bound and the checked statement remains relevant.

Popular right now

Roborock RockNeo Q110H Wire-Free Robot Lawn Mower

AmazonView Latest Price

iRobot Roomba Max 715 Pet Robot Vacuum with AutoEmpty Dock

AmazonView Latest Price

Dyson Car+Boat Cordless Handheld Vacuum

AmazonView Latest Price

As an affiliate, we earn on qualifying purchases.

From SAT Algorithms to ETH

In k-SAT, each rule places a limit of k on the number of variables, or their negations, combined in a clause; the task is to determine whether assignments can satisfy all clauses. With n variables, trying every assignment takes 2^n possibilities. Researchers have developed faster algorithms, but the historical improvements described in the source still have exponential running times. ETH, introduced by Impagliazzo, Paturi and Francis Zane in 2001, formalizes the conjecture that 3-SAT cannot be solved in subexponential time.

ETH is stronger than the familiar P-versus-NP question. P versus NP asks whether every problem whose proposed answers can be checked efficiently can also be solved efficiently. Refuting ETH would weaken a stronger assumption about the rate at which certain hard problems can be solved; it would not show that P equals NP. The Clay Mathematics Institute continues to list P versus NP as unsolved.

OpenAI presented the ETH work alongside two other research claims on October 26: a proof concerning local smoothing in harmonic analysis and a proposed existence result in quantum chromodynamics and a grand unified completion of the Standard Model. The source says the physics materials are being held back under a process led by Nobel laureate David Gross, while the computer-science preprint and code were released for review. OpenAI characterized the manuscripts as “a form of raw research data.”

Expert Review and Proof Scope

The central open issue is whether independent researchers will confirm the paper’s reasoning and its claimed implications. The review group is expected to examine the work, but no completed assessment is reported in the source material. Until that review is public, OpenAI’s description of the result remains a company-announced claim supported by a preprint and a machine-checked formalization.

There is also a specific scope question: the Lean code reportedly checks a weaker runtime bound than the preprint’s sharpest one. The available material does not explain in enough detail how the formalized result maps onto every claim in the paper, or whether the stronger bound will later be formalized. The proof-checking result should not be described as independent confirmation of claims outside the formal statement.

The source report also notes that an earlier OpenAI release of AI-generated manuscripts led to withdrawals and revisions after errors were found. That history does not establish a problem with this proof, but it underlines why review matters. The status of any practical algorithmic benefit is also unknown; no near-term speed-up for commercial systems or everyday computing is established.

Reviewers Test the Formal Claim

The next milestone is the external group’s examination of the preprint and its Lean formalization. The source material identifies the group’s leads as Russell Impagliazzo, Ramamohan Paturi and Ryan Williams, but does not give a completion date or report findings. Their review will need to clarify both whether the mathematical argument holds and how far the machine-checked statement supports the paper’s headline claim.

OpenAI says it has agreed to fund a conference on the result. A public review, revisions to the paper or formalization, and explanations of any differences between the two runtime bounds could help establish the claim’s standing. For now, the result is best treated as a significant AI-produced research claim under review, not as a settled change to computing practice.

Key Questions

What is OpenAI claiming its model proved?

OpenAI says an internal AI model produced a proof of a deterministic, subexponential-time algorithm for exactly counting solutions to fixed-width k-SAT formulas. For 3-SAT, the claim contradicts the Exponential Time Hypothesis, if the proof and its stated implications withstand review.

Does this mean P equals NP?

No. The claimed algorithm remains slower than polynomial time. Refuting ETH would challenge a stronger assumption about the difficulty of 3-SAT, but it does not settle the P-versus-NP problem, which remains open.

What does the Lean check confirm?

Lean checks whether the formal proof encoded in its system follows from the specified rules and assumptions. According to the source report, the formalization covers a weaker runtime bound than the preprint’s sharpest claim, so the check should not be taken as confirmation of every statement in the paper.

Has the proof been independently verified?

Not yet, according to the supplied source material. A group led by Impagliazzo, Paturi and Williams is expected to examine the work. No completed independent review is reported.

Will this make computing faster soon?

No immediate practical speed-up has been established. OpenAI says practical improvements would require significant further research, and the source gives no evidence that current products or computing systems have become faster because of the result.

Source: ThorstenMeyerAI.com