Terence Tao takes the lead to declare war, and human mathematicians unite to boycott OpenAI

The situation has completely broken down.

Terrence Tao, a leading figure in the mathematics community and Fields Medal laureate, led the Association for Human Mathematicians (AHM) in releasing a landmark joint statement —

Full-scale boycott of OpenAI.

Releasing more than 700 documents at one go is not scientific research at all, but a pure demonstration of computing power hegemony.

We strongly urge all mathematicians to cease cooperation with OpenAI and return to the scientific vision centered on human comprehension.

Why have human mathematicians collectively stood on the opposite side of model companies overnight?

It is simply because OpenAI has gone too far.

Some time ago, after OpenAI claimed to have solved the Millennium Prize Problems, Terrence Tao joined 25 Fields Medal laureates to issue a joint appeal, hoping that commercial companies would slow down their pace.

It is absolutely crazy that AI companies are accelerating endlessly without any idea of what will happen afterwards.

We must slow down. There is absolutely no reason to go so fast, not a single reason at all.

As it turns out, the commercial giant didn't take it seriously at all...

OpenAI then directly released more than 700 machine-generated proof documents, which left many mathematicians around the world almost fainting in shock when they saw them after waking up.

In the words of theoretical computer scientist Scott Aaronson:

This is an apocalyptic catastrophe for the mathematics community, the Mathocalypse.

Now, the collective counterattack of the entire mathematics community has officially begun.

Association for Human Mathematicians Declares Open War

The organization led by Terrence Tao to take the lead in confronting commercial giants this time is called the Association for Human Mathematicians (abbreviated as AHM).

It is no ordinary research group, but an industry defense line spontaneously established by the world's top mathematicians in response to the impact of generative AI.

The core purpose of the organization is to safeguard human understanding of the essence of mathematics and prevent academic norms from being completely destroyed by unrestrained brute-force computing power.

The joint statement led by AHM is extremely stern in tone. It brings up past grievances as soon as it opens, directly hitting OpenAI's soft spot:

The company, which is currently embroiled in lawsuits over allegations of illegal plagiarism, copyright infringement and trademark dilution, has released a number of manuscripts claiming to contain solutions to several well-known problems in the mathematical community.

It then struck back unceremoniously:

No one asked you to solve these proofs.

Why is the entire mathematics community so furious?

After the plagiarism controversy over the Millennium Prize Problems escalated, OpenAI, in order to show its compliance stance, set up the Advisory Group on Mathematics and AI (AGMAI) together with the Institute for Advanced Study in Princeton.

At that time, the official side published a special announcement, claiming to carry out in-depth co-construction with the mathematics community.

However, the so-called close cooperation turned out to be nothing but a public relations facade, and the opinions of the expert group were completely ignored...

The first advisory report released by the expert group clearly issued a red line warning:

Cutting-edge AI companies should not test advanced mathematical problems on their internal models without prior authorization.

So what was the result?

OpenAI turned a deaf ear to it.

It immediately used its internal model to solve 8000 problems in one go, and casually threw 700 machine-generated drafts in front of everyone.

Every word in AHM's statement hits the nail on the head:

Ignoring the core premise of the advisory group's position shows that OpenAI has complete disdain for scientific research norms.

Mathematicians have a specific vision for progress, which is rooted in historical considerations and the inherent logic of the discipline itself.

We reject OpenAI's claim that this release has promoted the development of mathematics, and urge mathematicians and the public to maintain due skepticism about the value of this publishing model.

700 Proofs Stun Mathematicians

Where on earth did these more than 700 documents that broke the entire mathematics community come from?

Public data shows that OpenAI used its internal model to test about 8000 open mathematical problems, with a hit rate of around 5%.

On average, solving one high-difficulty problem only consumed 3 hours of GPT-Pro level computing power.

Just three hours of computing power can directly crush the research accumulation of a scholar for several years or even a lifetime.

Now that AI has evolved to this stage, even mathematicians standing at the pinnacle of human pure intelligence have truly felt the sense of powerlessness of being crushed by algorithms.

What is even more absurd is yet to come.

When scholars forced themselves to open these documents, they were completely stunned.

Scott Aaronson mentioned the experience of his wife Dana Moshkovitz.

Dana has devoted all her academic career to the top theoretical computer science problem, the Unique Games Conjecture.

However, the document released by OpenAI this time explicitly states that the conjecture has been proven.

After reading the document overnight, Dana felt nothing but absurdity and despair. She wrote in a text message:

This feels like something written by someone on hallucinogens.

The entire paper is full of extremely chaotic logical leaps, piles up a large number of a priori conclusions without any explanation of their applicability, and uses weird constructions that human beings have never seen before throughout.

The paper is so poorly written that it is completely unreadable without AI assistance. It is full of alien-like crazy logic.

There is no peer review, no readable human logic, let alone respect for and explanation of the work of predecessors.

AI giants directly threw a bunch of unintelligible drafts to the academic community, trying to get the world's top scholars to work for them for free as human problem-verifying tools.

The "proof indigestion" that Terrence Tao has been warning about before is breaking out in full swing.

Who Is Killing Mathematics?

Is proving difficult problems all that mathematics is about?

This is a proposition that Terrence Tao has repeatedly questioned.

In the traditional mathematics community, when a major conjecture is solved by humans, it will give rise to a large number of seminars, lectures and cross-disciplinary cooperations, which will eventually become classic content in textbooks and inspire an entire generation of young successors.

The situation is completely different now.

The person who pressed the enter key does not understand the proof at all, while the mathematicians responsible for reviewing the proof are tortured to agony by the unintelligible content.

Many young scholars are even afraid to make public the directions they are exploring for fear of being preempted by large models.

Terrence Tao calls this kind of damage irreversible pollution:

Once a problem is claimed to be solved, it can never go back to its unsolved state. Even just knowing that the answer exists will pollute other research paths that may lead to deeper insights.

Commercial computing power is industrially harvesting mathematical problems, leaving human scholars with a barren land that they cannot understand.

As early as 2024, Terrence Tao envisioned industrialized mathematics.

At that time, Terrence Tao was very optimistic, firmly believing that humans are still the dominators, and AI can at most assist mathematicians just like a chess engine.

No one could have expected that the situation of offense and defense has been completely reversed in just a short period of time.

AI is responsible for solving difficult problems that are out of human reach, while humans have been reduced to tools responsible for checking calculations and sorting out notes...

When a few hours of machine operation can easily surpass a scholar's lifelong exploration, this ancient and pure discipline is being pushed to an unknown crossroads.

One More Thing

Facing the fierce conflict between AI giants and the mathematics community, when many people lament that mathematics is dead, Turing Award laureate Yann LeCun also came forward to comment.

But his attitude seems to be completely different from that of star mathematicians such as Terrence Tao:

On the contrary.

Mathematics is ushering in a new era. Formal proofs will be largely automated, and the focus will shift to developing new concepts, new abstractions, new definitions and new conjectures.

The invention of ships reduced the importance of swimming, but allowed us to discover new continents.

Reference Links:

[1]https://terrytao.wordpress.com/2026/10/07/ahm-statement-on-openais-october-6-release-of-mathematical-documents/

[2] https://the-decoder.com/subscription/

[3] https://x.com/ylecun/status/2108118582856925198

This article is from the WeChat official account "QbitAI", written by Jay, and is published with authorization from 36Kr.