The Difference Between OpenAI’s AI Claim And What The Proof Establishes

🔍 Read the full analysis: The Difference Between OpenAI’s AI Claim And What The Proof Establishes on ThorstenMeyerAI.com

Buying for a business?Offer from Amazon

Get business pricing on office and shipping supplies

• 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 internal AI model had produced a proof of an algorithm that would refute the Exponential Time Hypothesis (ETH), a long-standing conjecture about the difficulty of solving certain logic problems. The preprint and a Lean formalization are public, but independent expert review has not been completed, and the formalization checks a weaker runtime bound than the paper’s sharpest claim.

The claim concerns k-SAT, a problem asking whether truth values can be assigned to variables so that every rule in a formula is satisfied. OpenAI’s preprint describes a deterministic algorithm that exactly counts solutions for fixed k in subexponential time and polynomial memory. For 3-SAT, the authors say this would contradict ETH, which holds that solving 3-SAT requires exponential time.

The preprint also says the result refutes the stronger Strong Exponential Time Hypothesis and three related hypotheses. Those are claims made by the paper, not conclusions independently certified by reviewers. The supplied source says the paper’s sharpest runtime bound is not covered by the Lean formalization; Lean checks a slightly weaker bound.

OpenAI released the preprint and Lean code after consultation with outside researchers. The source author counted 3,977 files and about 710,000 lines in the formalization, while noting that the central formal statement is short. OpenAI says its system found the disproof on October 16 and the formalization followed the next day. The preprint is dated October 17 and was revised October 25.

What an ETH Refutation Would Change

If the result survives review, it would revise a widely used assumption about the limits of algorithms for hard computational problems. ETH has supported conditional lower-bound arguments: researchers use it to show that certain tasks are unlikely to have substantially faster algorithms. A valid counterexample could require some of those results to be reconsidered, depending on their exact assumptions.

That does not mean practical software will become faster immediately. The claimed algorithm remains slower than any polynomial-time algorithm, and OpenAI says practical improvements would require significant further research. The source author’s arithmetic also indicates that the speed-up grows very slowly with problem size. The result is primarily a development in theoretical computer science and in how mathematical claims made with AI might be checked.

It is also not a solution to P versus NP. That question asks whether every problem whose answer can be checked efficiently can also be solved efficiently; it remains open. ETH is a stronger assumption about computational hardness, so refuting it would weaken that assumption without resolving P versus NP.

Products Worth Considering

Logitech MX Mechanical Wireless Illuminated Keyboard, Tactile Quiet, Graphite

AmazonView Latest Price

Logitech MX Mechanical Full-Size Wireless Mechanical Keyboard, Tactile Quiet Switches, Backlit Keys, Bluetooth, USB-C, Multi-Device, Graphit

AmazonView Latest Price

Logitech MX Mechanical Mini Mac Wireless Keyboard Low-Profile – Space Grey

AmazonView Latest Price

As an affiliate, we earn on qualifying purchases.

The Hypothesis Behind the Claim

In k-SAT, each rule contains at most k variables or their negations, and the task is to find an assignment that makes all rules true. With n variables, direct enumeration considers 2^n assignments. Researchers have developed faster approaches, but the known examples in the source still take exponential time. ETH, formulated by Russell Impagliazzo, Ramamohan Paturi and Francis Zane in 2001, asserts that 3-SAT cannot be solved in subexponential time.

The source describes ETH as influential but cautions against presenting it as universally accepted. It quotes computer scientist Scott Aaronson’s 2017 survey saying there was “by no means a consensus in the field that ETH is true.” That distinction matters: a result can challenge a widely used research assumption without overturning a proved theorem.

OpenAI presented the result alongside announcements in mathematics and physics. It says outside experts, described as “shepherds,” advised on how each project should be released. For the computer-science claim, an external group is expected to explain and verify the work. The account of the release process and timeline comes from OpenAI’s own announcement.

What Review Has Yet to Establish

The main unresolved issue is whether the preprint’s argument holds up under independent expert scrutiny. The source says a group led by Russell Impagliazzo, Ramamohan Paturi and Ryan Williams will examine it; Impagliazzo and Paturi were two of ETH’s original authors. The review’s findings and timetable are not yet provided.

There is also a gap between the paper and the machine-checked result: Lean formalizes a weaker runtime bound than the sharpest one claimed in the preprint. A formal checker can verify that a formalized proof follows from its definitions and rules, but that does not by itself establish that the formal statement covers every headline claim in the paper. The source reports that the Lean code was scanned, but does not describe a completed independent audit of the formalization.

The full consequences for existing complexity results remain uncertain until the claim is reviewed and its exact assumptions and bounds are assessed. No practical performance gains or revised algorithms are established by the announcement alone.

Independent Review and Formal Checks

The next key step is the external group’s examination of the preprint and its relationship to the Lean formalization. Readers will need to see whether reviewers accept the proof, identify errors, or request revisions, and whether the sharpest runtime claim can be formalized and checked at the same level as the weaker bound.

OpenAI says it has agreed to fund a conference on the result. The source does not provide a date or say when the reviewers will report. Until that work is public, the appropriate description is that OpenAI has announced an AI-produced proof and released supporting materials—not that the refutation has been independently established.

Key Questions

What is OpenAI claiming its AI proved?

OpenAI says an internal model produced a proof of a deterministic algorithm that exactly counts solutions to fixed-k SAT formulas in subexponential time and polynomial memory. For 3-SAT, the preprint says this contradicts ETH.

Has the proof been independently confirmed?

Not according to the supplied source. Lean has machine-checked a formalization, but an external expert group’s review is still pending. The formalization checks a weaker runtime bound than the paper’s sharpest claim.

Does this solve P versus NP?

No. The result concerns ETH, a hypothesis about the hardness of 3-SAT. P versus NP remains an open problem, and the claimed algorithm is still slower than polynomial time.

Will the result make computers solve hard problems faster now?

No immediate practical improvement is established. OpenAI says applying the result will require significant further research, and the source describes the algorithm’s speed-up as growing very slowly with problem size.

What does Lean checking establish?

Lean checks that a formalized statement follows from the definitions and proof steps encoded in the system. In this case, the source says the formalization covers a weaker bound than the paper’s sharpest runtime claim, so the two should not be treated as identical.

Source: ThorstenMeyerAI.com