OpenAI’s AI May Challenge The Exponential Time Hypothesis—Here’s What’s Known

🔍 Read the full analysis: OpenAI’s AI May Challenge The Exponential Time Hypothesis—Here’s What’s Known on ThorstenMeyerAI.com

Before you orderOffer from Amazon

Get audio and creator gear delivered free with Prime

• Fast, free delivery on millions of items
• Prime Video, Amazon Music and more included
• Member-only deals all year

Start your free Prime trial →Free trial for eligible customers · Cancel anytime

As an affiliate, we earn on qualifying purchases.Hide

OpenAI says an unreleased AI model has found an algorithm that would refute the Exponential Time Hypothesis (ETH), a central conjecture about the difficulty of solving Boolean satisfiability problems. The company released a preprint and Lean proof code on October 26, but the result has not yet been independently reviewed, and the formalized statement checks a weaker runtime bound than the sharpest claim in the paper.

The preprint describes a deterministic algorithm that exactly counts solutions to any k-SAT formula, for fixed k, in subexponential time and polynomial memory. For 3-SAT, such a result would contradict ETH, which holds that 3-SAT requires exponential time. The authors also say the work refutes the stronger Strong Exponential Time Hypothesis and three related hypotheses.

OpenAI says its system found the disproof on October 16, and that the Lean formalization followed a day later. The preprint is dated October 17 and was revised October 25. The source article reports that the public Lean repository contains 3,977 files and about 710,000 lines of code, while the formal statement itself is much shorter.

There is a material difference between the paper and the checked code: the Lean formalization verifies a slightly weaker runtime bound. The sharpest bound in the preprint is not covered by the formalization, according to the source. OpenAI has asked an outside group led by Russell Impagliazzo, Ramamohan Paturi and Ryan Williams to examine the computer-science result. Impagliazzo and Paturi were among ETH’s original authors.

What an ETH Refutation Would Change

If the result withstands review, it would overturn a widely used assumption about the limits of algorithms for NP-complete problems. ETH has supported conditional hardness results across theoretical computer science: researchers use it to argue that particular problems are unlikely to have substantially faster algorithms. A refutation would require those results and their assumptions to be reassessed.

That would not mean ordinary computing tasks suddenly become fast. The stated runtime is roughly 2 to the power of n divided by the square root of log n, still superpolynomial and potentially impractical for large inputs. OpenAI itself says practical improvements would require significant further research. The importance for now lies chiefly in what the result could mean for theory, and in the method used to produce and check it.

The episode also tests a possible research workflow: an AI system generates a proof, formal verification software checks a precise statement, and specialists independently inspect the argument. A machine-checked proof can catch errors relative to its formal statement, but it does not resolve whether the formal statement matches every claim in a paper or whether the work has broader consequences. Those questions remain with human reviewers.

Products Worth Considering

Bose QuietComfort Wireless Noise-Cancelling Over-Ear Headphones, Black

AmazonView Latest Price

Sony WH-1000XM6 Wireless Noise-Canceling Over-Ear Headphones, Blue, Bundle

AmazonView Latest Price

Apple iPad Air (2025) 13‑inch Wi‑Fi 256 GB – Space Gray (M3 chip)

AmazonView Latest Price

As an affiliate, we earn on qualifying purchases.

From SAT Algorithms to ETH

In k-SAT, each rule involves at most k variables, and the task is to determine whether there is an assignment of true or false values that satisfies all rules. With n variables, checking every possible assignment takes 2ⁿ combinations. Researchers have developed faster-than-brute-force algorithms, but those still take exponential time.

ETH, proposed by Impagliazzo, Paturi and Francis Zane in 2001, formalizes the expectation that 3-SAT cannot be solved in subexponential time. It is stronger than the P-versus-NP question: refuting ETH would not show that P equals NP, nor settle whether every efficiently checkable problem can be solved efficiently. P vs. NP remains open.

This is part of a broader OpenAI release of AI-generated research. The company announced three results on October 26: the ETH claim; a mathematics result concerning Sogge’s local smoothing conjecture; and a physics claim concerning quantum chromodynamics and a proposed completion of the Standard Model. The source reports that an earlier release of more than 700 manuscripts was followed by corrections: three were withdrawn over a sign error and 14 revised. That history underscores why formal checks and independent review matter.

What Reviewers Still Need to Check

The main open question is whether the paper’s argument is correct and whether its conclusions follow at the strength claimed. The independent group has not yet signed off. The source says the Lean proof covers a weaker bound than the paper’s sharpest runtime, leaving a gap between what the formal code checks and the strongest headline claim.

It is also not yet clear how reviewers will assess the relationship between the preprint, the formal statement and the full Lean code, or when their examination will be complete. The source material does not provide a final review timetable or report. Claims that the result refutes several additional hypotheses likewise remain claims in the preprint until experts have examined them.

The extent of practical consequences is also unknown. Even if the theoretical result is validated, the stated algorithm remains superpolynomial, and no immediate application or engineering improvement has been established.

Independent Review and Formal Checks

The next milestone is the outside group’s examination of the computer-science result. Its members are expected to explain and verify the work; the source material gives no date for a conclusion. OpenAI has also agreed to fund a conference on the result, providing a venue for further discussion.

Readers should watch for a review that distinguishes the paper’s strongest runtime claim from the weaker bound currently covered by Lean. Any eventual confirmation should make clear which statement has been verified, what assumptions it relies on and whether the formalization is extended to cover the sharper result. Until then, the ETH refutation is a consequential unverified research claim, not a settled change to computer science.

Key Questions

What is OpenAI claiming its AI proved?

OpenAI says an internal model found a deterministic, subexponential-time algorithm for exactly counting solutions to k-SAT formulas for fixed k. Applied to 3-SAT, the claim would refute the Exponential Time Hypothesis.

Has the result been independently verified?

No. Lean code has checked a related, slightly weaker runtime bound, but an outside group led by Russell Impagliazzo, Ramamohan Paturi and Ryan Williams has yet to complete its review.

Does this solve P vs. NP?

No. ETH is a stronger hypothesis than P ≠ NP. Refuting ETH would not establish that P equals NP, and the P-versus-NP problem remains open.

Will computers become faster because of this result?

There is no confirmed immediate practical speed-up. The claimed runtime remains superpolynomial, and OpenAI says practical improvements would require significant further research.

Source: ThorstenMeyerAI.com