For decades, solving the world’s most difficult mathematical problems has been considered a uniquely human endeavor, requiring imagination, perseverance, and the ability to explore ideas beyond established methods.
Meanwhile, AI is largely viewed as a tool for calculation and analysis. A Harvard mathematician’s collaboration with an experimental model of artificial intelligence gained widespread attention after the system helped identify a potential solution to an 87-year-old mathematical puzzle. Although the result awaits formal verification, researchers are studying whether artificial intelligence could become a valuable partner in advanced mathematical research.
An AI model may have solved an 87-year-old mathematical problem
As arXiv reported, the problem at the heart of the debate is the Jacobian conjecture, first proposed in 1939 by German mathematician Eduard Ott-Heinrich Keller. Despite its simple appearance, it has withstood decades of serious mathematical effort and remains one of the most famous unsolved questions in algebra.In essence, the estimation concerns some polynomial transformations. In simple terms, it asks whether it is always possible to invert a certain type of mathematical map using another polynomial graph.
The statement is easy enough to write down, but proving or disproving it has defeated specialists for generations.Sometimes, proving a mathematical statement wrong doesn’t require a complex argument. One valid counterexample is sufficient. If even one example violates the rule, the guess is no longer valid.In one of the X posts, mathematician Levent Albog said: “Hey, the Jacobian conjecture is wrong.
Levent Alboge: A mathematician from Harvard University who worked with the anthropic AI model Fable
Levent Albaugh, a 33-year-old Harvard mathematician, recently decided to tackle this question with the help of Anthropic’s experimental artificial intelligence model known as Fable.Working over the course of a single Sunday, Alpöge specifically looked for a counterexample rather than a complete proof. According to posts shared online, the collaboration has produced a candidate example that appears to contradict long-standing conjecture.The proposed counterexample itself is remarkably compact, consisting of only 216 characters. Its size has surprised many observers because the difficulty of the problem is not in writing a long expression, but in finding an expression that really satisfies all the mathematical requirements.
Artificial intelligence and mathematicians verify basic calculations before formal review
This result has not yet passed formal peer review, which remains an essential part of mathematical research. Independent verification is still needed before the result can be accepted as final.However, early reactions have attracted attention. Several mathematicians have reportedly independently examined the calculations and arrived at the same conclusion regarding the calculation in question.
Others have reproduced parts of the work using established mathematical software, including SymPy, while formal proof tools such as Lean have also been used to test aspects of the argument.Additional AI systems have also been integrated into the verification process. Various large linguistic models were asked to examine the proposed counterexample, creating an unusual situation where an AI-assisted mathematical result was verified with the help of other AI systems alongside traditional methods.
A different role for artificial intelligence
Large linguistic models are gradually finding a place within mathematical research, but not in the way many people initially envisioned. They are rarely expected to replace mathematical reasoning. Instead, researchers have begun to use them as collaborators who can suggest approaches, generate examples worth investigating, or identify gaps in the discussion that deserve greater attention.This changes the pace of research rather than the criteria by which results are judged.
Every significant claim still requires careful verification, but some exploratory work that used to take weeks can now be done much more quickly.Formal verification has become another area where AI is starting to make a practical contribution. Systems capable of translating mathematical arguments into languages understood by proof assistants allow complex logic to be examined mechanically, reducing the chance of subtle logical errors remaining unnoticed.
A final ruling awaits as the role of artificial intelligence in mathematics continues to grow
Whether the counterexample proposed by Alpöge will ultimately survive detailed scrutiny remains uncertain. Mathematical history contains many examples of popular claims that were later corrected or retracted after close examination.Even if further review identifies problems, this incident highlights something that many researchers have noted over the past few years. AI systems are becoming increasingly useful during the creative stages of mathematical work, helping researchers explore possibilities that previously took much longer to test.
