AI kills the Jacobian conjecture - the details

[00:00:00]Imagine a universe made of fluid rubber.

[00:00:03]You can stretch it, bend it, and morph it in any direction. As long as you never tear the fabric or crease it into a sharp fold, the space remains intact.

[00:00:13]Otto Keller's 1939 Jacobian conjecture proposed that if a space follows a strict local rule, a constant non-zero Jacobian determinant, it should never collapse globally. Under this rule, transformations are bijective. Every destination corresponds to exactly one starting point, leaving no overlaps and no holes. This geometric intuition, that local stability implies global stability, remained an unproven assumption for 87 years, becoming one of the most persistent open problems in algebraic geometry.

[00:00:49]For decades, researchers attempted to find a universal proof. The goal was to demonstrate that this no folding rule held true for every possible polynomial equation across any number of variables.

[00:01:01]Proving the conjecture is difficult because it applies to high-dimensional complex planes. These are mathematical structures that obey logical rules, but cannot be represented in three-dimensional physical space, making them impossible to verify through simple visual intuition.

[00:01:18]The history of the problem is marked by a cycle of error. Mathematicians frequently published what appeared to be complete proofs, only for colleagues to discover subtle logical gaps months or years later.

[00:01:31]By the 2020s, the field had reached a stalemate. Without a proof or a counterexample, it was unclear if the rule was a universal law or if the anomaly required to break it was simply too specific for manual discovery.

[00:01:45]On July 20, 2026, mathematician Levan Alpoge posted a screenshot of three complex equations to social media. While the community expected a lengthy theoretical paper, the breakthrough was a concrete polynomial map in three variables.

[00:02:01]Alpoge's map, using variables P, Q, and R, is a precise set of instructions for moving points between high-dimensional coordinates. Calculating the Jacobian determinant for these specific equations yields a constant result, exactly -2.

[00:02:17]Because -2 is never zero, the map obeys Keller's local rule. It never collapses dimensions in any local neighborhood. Up close, every part of this transformation looks perfectly legal and reversible.

[00:02:31]However, when these equations are applied globally, the transformation experiences a collision.

[00:02:38]Plotting three distinct points, (0, 0, -1/4), (1, -3/2, 13/2), and (-1, 3/2, 13/2) on this grid, shows the collision.

[00:02:50]The first moves to output (-1/4, 0, 0), but the equations drag the second and third points to the exact same location.

[00:02:59]Three different starts yield one identical destination.

[00:03:03]This means the map is not injective.

[00:03:06]Despite following the local rules, the space has folded over itself. By producing the same output from multiple inputs, the equations prove that a global inverse is impossible. A single overlapping point in a three-variable system is enough to disprove the conjecture.

[00:03:23]One specific counterexample has invalidated the search for a universal proof.

[00:03:29]This failure in three dimensions has a cascading effect. Using a technique called identity padding, mathematicians can use Alpoge's map to show that the conjecture is also false in four, five, and all higher dimensions.

[00:03:43]Alpoge and his colleague Akil developed the approach, but they used an AI system named Claude Fable to help navigate and check the complex polynomial structures.

[00:03:53]The work took place during the World Cup final with the AI assisting in the symbolic manipulation required to ensure that Jacobian stayed constant while the global points collided.

[00:04:04]Once the equations were shared, researchers used symbolic computation software to verify the results, providing the first definitive evidence that the conjecture was false.

[00:04:14]The resolution of the Jacobian conjecture provides a new data point for algebraic geometry. Researchers can now study the specific properties of these folding maps to understand why the local to global transition fails.

[00:04:27]This discovery highlights a shift in how mathematics is conducted. The AI functioned as a specialized lens scanning high-dimensional possibilities to pinpoint an anomaly that had escaped decades of manual proofreading.

[00:04:41]By delegating the symbolic verification to an algorithm, Alpo Ge found the specific coordinate where the math breaks.

[00:04:48]In modern geometry, breakthroughs are increasingly found at the intersection of human hypothesis and algorithmic search.