OpenAI attacks 722 mathematical puzzles overnight, and Liman's guess has been proven.

  It's the first time in the history of the world.

OpenAI is making people angry again!

Today, the world of mathematics and AI is shocked by this news.

Without warning, without peer review, and even ignoring the long-standing courtesy of academia, OpenAI has released a series of new mathematical results resulting from an internal front-end model.

They're just ruthlessly taking it out of the GitHub project library.math 。

It contains 722 mathematical manuscripts covering 372 previously unresolved top-level mathematical difficulty families.

Among them, the unpublished AI model of OpenAI proves Liman's conjecture and simultaneously releases Lean formalized verification. If confirmed, this would be a historic breakthrough in the field of mathematics and a milestone in the history of AI development!

According to OpenAI's disclosure, proof of the vast majority of problems is that it only takes an average of 3 hours for ChatGPT Pro to think for an internal model they haven't released!

Ottman X: We are entering a whole new era of discovery

This has angered mathematicians.

This list of problems is suffocating.

OpenAI and mathematicians have long been at the forefront of this "academic massacre".

As early as August of this year, OpenAI secretly convened 40 of the world's top mathematicians to raise a suffocating topic: "What if AI completely surpasses humans in pure mathematics?"

At the time, OpenAI secretly revealed that its internal model had broken through hundreds of pending cases.

Bryna Kra, a renowned mathematician at Northwestern University, recalled that the atmosphere was "extreme excitement and extreme fear".

Scholars are urging OpenAI: Don't just tweet or blog like Red Hat, but publish rigorous papers according to academic standards, giving humanists time to digest and validate them.

However, OpenAI claims sovereignty in the roughest possible way, even before the academic breakthrough of the Navi-Stokes equation.

Nestor Guillen, a visiting professor at New York University, has angrily sued

In the eyes of mathematicians, the actions of these AI giants are just like gangs! There is extreme panic, not only because of AI itself, but because of the intellectual power of the highest dimensions of humanity, which is being ruthlessly monopolized by a very small number of tech oligarchs.

It has been revealed that some of the OpenAI engineers have privately agreed: "Classical mathematics is dead today, and AI will inevitably end the careers of most professional mathematicians".

One of the most prestigious awards in statistics, the winner of the "Copps President's Award", OpenAI researcher Suzanne Jeffery said: "It's like the beginning of a general paradigm shift in the human understanding of intelligence, like the Copernican paradigm".

Of all the fortresses that were captured, the first to drive the entire mathematical community crazy was the result of a super-outcome numbered Result 003 that ripped open the door to the mathematical world's ultimate Holy Grail of Liman's conjecture.

Lyman's conjecture is the mathematicalally recognized "pearl on the crown", and hundreds of theories in modern mathematics are all based on the basis of Lyman's conjecture.ζ(sAll the unusual zeros are in real R.sFor more than 160 years, it has been difficult for humans to even rule it out at zero points away from the 1/2 region.

Furthermore, a phantom of a "Rondo-Siegler zero-point" lurks in which some Dickey-L-functions may have anomalous zero-points on the real-time axis very close to 1, which hinders hope.

In this open-source manuscript, the OpenAI model completely conquered the "pre-Rhyman conjecture": proving all of the DickeyL-functions, throughout the core R.sIn the half-plane of 7/8, there are absolutely no zeros!

Moreover, it completely wiped out the Rondo-Seagull Zero.

OpenAI admits in its GitHub explanation that the vast majority of problems were run entirely automatically by the model, and the research team conducted extremely rigorous artificial scrutiny and readability in only the zero-point area of the Liman Zeta function.

It's not even close to the final R.s= 1/2, but pushing the zero-point area to a fixed constant boundary (7/8 and 11/12), and consistency excluding the Siegel zero, has been a surprise leap in parsimony for half a century!

In the field of computer science, if P vs NP is the ultimate crown, then the "ordinary NP-difficulty below the fundamental semi-definite threshold" is the "King of the Uncrown" that determines the limits of human algorithms.

This is also one of the most disruptive studies in the OpenAI results database (number Result 102).

What is NP-Hard?

In the real world, large-scale optimization problems such as chip wiring, logistics scheduling, route planning, coloring are all classified as NP-Hard problems.

Human beings do not figure out the best solution in multi-dimensional time, they can only retreat and seek the next approximate solution.

In 2008, computer scientist Prasad Raghavendra published a landmark paper showing a surprising conclusion: for any fixed finite constraint language (Max-CSP), the approximate ratio between Basic-SDP is the absolute limit for the theory of multi-time algorithms!

However, this great theorem has a deadly premise that it must be built on the foundation of the "only game guess" (UGC).

UGC is a problem of the century proposed by Subhash Khot in 2002.

If the UGC were false, Raghavendra's theoretical pyramid would collapse instantly, which was the "architectural pyramid" of theoretical computer for nearly 20 years.

Over the past two decades, the dream of countless theoretical computing scholars has been to prove that the Basic-SDP threshold gap problem itself is NP-Hard, in a pure, unconditional, classical framework based solely on P≠NP.

If this conclusion is true, it would mean that under the pure P≠NP assumption, any algorithm of polynomial time-determination that attempts to exceed the Basic-SDP representation would be impossible in mathematical logic!

How does AI actually tear this barrier apart?

In the first step, AI first reviewed Raghavendra's original framework, confirming that duplication variables and local probability distributions could not provide loopholes to construct countercases.

AI realizes that the most central obstacle to bypassing UGC is that in the classic PCP (probability verifiable) construct, volumes indicate a "leakage" of projection coordinates, which makes it easy for scammers to access.

In order not to undermine integrity while suppressing information leaks, AI abandoned the smooth function path and introduced an algebraic core over a finite domain of characteristics:

AI then designed non-linear decoders with variables such as shift positions.Extreme sensitivity to small noises can be constantly captured by high-level linear characteristics, which resolves the problem of information leakage.

Then, AI uses extremely rare projections of probability only, combined with innovative "linear fiber richness reasoning", while maintaining a rich decoding coordinate that quickly reduces statistical errors to zero and completely blocks the possibility of fraud on local slices.

Ultimately, AI divides the entire macro proof into two precise phases:

The first step: unconditional construction of Unique Games with almost perfect completeness ((1−ε) with arbitrary low reliability ((δ) difficulty;

The second stepAccess the authoritarian test system, using the low-impact Gaussian variable substitution, to pass the gap losslessly to an arbitrarily limited Basic-SDP threshold.

As a result, for the first time, AI is based purely on standards, on the premise that it is completely separate from UGC.P≠NPThe Basic-SDP threshold is established by the ordinary NP-hardness, which completely locks the theoretical physical boundaries of effective human approximation algorithms!

In the Result 01 manuscript, AI conquered a major fortress of Hodgkin's conjecture: a comprehensive proof of Abel's conjecture with multiplication of CM in all dimensions and excess dimensions!

The OpenAI official explains specifically:

Most of the results were automatically run by the standard model, but the proof of the Abel Cohn-Hoch conjecture is a special breakthrough in breaking conventional processes.

Not only that, but AI also extended the result to the arbitrarily finite multiples of the projection-replicating K3 curve, and demonstrated sequentially all of Abel's Tate conjectures on the finite field, as well as the Hodge standard conjecture under arbitrary characteristics.

The logic behind AI is as follows:

1. Transformation and projectionThe central difficulty in proving Hodgkin's conjecture is that abstract "Hodgkin classes" are inherently "algebraic". AI does not attempt to attack all fluids directly, but instead focuses on highly symmetrical CM Abellian and K3 curves.

2.KugaSatake corresponding algebraAI uses the deep-seated Kuga-Satake reaction to embed the K3 curve's transversal consonant into the second phase of the Abel consonant. It successfully demonstrates that the reaction itself is induced by a rational algebraic ring.

3. Degradation and diffusionSubsequently, AI used the symmetry of the Li algebra with the differential rigidity of the Hodgkin general point to construct an algebraic path from a special curve covering the whole of the self-symmetrical curve, proving that these Hodgkin classes must be completely enclosed in a rational domain by the closed algebraic chain.

This step, the equivalent of Hodge's guess on the face of it, is tearing up a huge gap!

In addition, OpenAI's manuscripts also contain a number of groundbreaking breakthroughs in the fields of quantum geometry and analytical geometry.

The normal two-point correlation of the multiplication function (result 007)

This is the central question in mathematics, involving the famous Chowla conjecture and Elliott conjecture. The central question is whether a finite multiple function tends to have a multiplication average of 0 under different planes.

AI demonstrates the usual two-point Chowla conjecture and saves the error of several secondary levels on each scale.

It's the same as the general Mallet conjecture (Result 087).

Mahler's conjecture in convex geometry has remained unresolved for decades.

It asserts that in n-dimensional space, the volumetric multiplication of a convex body and its polarity is obtained at the simplest form (for general convex bodies) or cube/cross polygon (for symmetric convex bodies).

AI solves both symmetrical and asymmetrical geometric Maller conjectures at the same time in all dimensions, and gives Hanner polynomials and polynomials the classification conditions for the formation of all equivalents.

After seeing this, the mathematical circle felt a profound sense of powerlessness and shock.

Previously, we thought that AI proved mathematics by just matching patterns in a lot of content.

But the manuscripts published today are full of "intuitive transfers", "constructive counterexamples", "Laplastic openings" and "physical intuitions" (such as heat flow simulations, the Hamilton system) that humans can understand.

Not only did it learn the human mathematical framework, it also created its own mathematical intuition.

Back to the weighty question at the beginning of the article: what should human mathematicians do when faced with a model that can produce top-notch results in an average of three hours?

Bryna Kra said: "We have to adapt in this area. It changes the way we operate, but it's also a time when we can look further ahead... It's a scary time, but it's definitely an extremely exciting time".

The era of classical mathematics may be over when the answers to more than 100 unresolved centuries-old puzzles lie in GitHub's codebase like a freezing stream of data.

However, humanity has only just begun its search for truth. In a new era in which the search for truth is engineered by artificial intelligence, humanity has just made its way to the starry ocean.

Tonight, the world of mathematics is destined to put no one to sleep.