AI Cracks Century-Old Math Problem
· news
Mathematicians Grapple with a ‘Very Rapid and Very Unsettling Change’
The latest milestone in AI’s progress through mathematics has left many mathematicians feeling like they’re watching a familiar scene play out in slow motion: another century-old problem falls, and with it, their grip on an increasingly automated profession. On Sunday, an Anthropic model solved the Jacobian conjecture, a puzzle that had puzzled German mathematician Ott-Heinrich Keller since 1939.
This breakthrough is part of a larger trend – AI’s stunning progress in mathematics since mid-2025, when OpenAI’s model first solved five out of six problems at the International Mathematical Olympiad. Since then, we’ve seen a string of high-profile successes: the Erdős conjecture on combinatorial geometry was disproved in May, and the Leiden Declaration on Artificial Intelligence and Mathematics was published in June, cautioning against AI reshaping mathematical knowledge without guardrails.
The Jacobian conjecture’s downfall is particularly striking. It concerns “maps” – a fundamental concept in mathematics – and the conditions under which you can determine an input given a set of outputs. The problem has been a thorn in mathematicians’ sides for decades, but Alpöge’s result offers a tantalizing glimpse into AI’s potential to tackle problems that have eluded human minds.
The impact on mathematicians is multifaceted. As Kevin Buzzard noted, the “how” without the “why” is a recurring issue when it comes to AI-driven breakthroughs in pure mathematics. This raises concerns about the profession’s future, as current models excel at producing correct answers but struggle to provide underlying reasoning or narrative.
Mathematicians have long argued that what sets them apart from machines is not just their computational prowess but their capacity for reasoning, explanation, and communication. Michael Harris wrote in Boston Review that the AI industry views reasoning as commercially worthless, treating human mathematicians like a “beta version” of intelligence. However, this perspective neglects the fundamental value of mathematics as an exercise in unalienated labor – a field where people can earn a living by playing with ideas, exploring patterns, and uncovering new truths.
The ongoing decline in federal funding for mathematics research is a stark reminder that mathematics is not immune to budget cuts. With PhD admissions at top universities on the rise, it’s clear that mathematicians are still being drawn to this field – but what happens when AI starts solving problems that once required human ingenuity?
Perhaps the real question we should be asking isn’t “Can machines do math?” but “What kind of math do we want our machines to do?” Do we prioritize efficiency and speed, or do we value the slow, deliberate process of mathematical discovery that has driven human progress for centuries? The answer will determine not just the future of mathematics but also the kinds of problems we choose to solve – and the kind of world we want to create.
As AI continues to write its own history in the annals of mathematics, one thing is certain: the profession will be forever changed. But if mathematicians are smart, they’ll focus on what makes them unique – their ability to convey meaning, context, and understanding through stories that connect human experience with mathematical truths.
Reader Views
- ADAnalyst D. Park · policy analyst
The Jacobian conjecture's fall is just the latest manifestation of AI's incremental encroachment into mathematicians' domain. While AI's successes are undeniably impressive, we must consider the practical implications of ceding mathematical problem-solving to machines. What happens when these models are applied in real-world contexts, where human intuition and context-driven reasoning are essential? The "how" without "why" is a recipe for black-box decision-making, which could have far-reaching consequences in fields like cryptography and optimization. As we accelerate down this path, it's crucial to prioritize not just the speed of breakthroughs but also the transparency and interpretability of AI's outputs.
- EKEditor K. Wells · editor
The Jacobian conjecture's fall is just the tip of the iceberg in AI's encroachment on mathematical territory. What we're not being told is how this will change the way students learn math. Will curricula shift to focus on understanding the 'why' behind AI-driven solutions, or will educators be forced to play catch-up with the latest models? The math community needs a clearer plan for integrating human insight into AI-generated results – not just in research papers, but in classroom materials and textbooks too.
- RJReporter J. Avery · staff reporter
The rapid ascent of AI in mathematics is more than just a novelty - it's a symptom of a deeper issue: our profession's failure to adapt its pedagogy to the changing landscape. For decades, we've taught students how to solve problems using manual reasoning and intuition, but this approach won't cut it when machines can produce answers without explanation. It's time for math education to shift focus from 'how' to 'why', ensuring that future mathematicians not only understand algorithms but also comprehend the underlying principles that govern them.