AI Claimed a Solution to the 167-Year-Old Riemann Hypothesis, While a Human Released a Competing Proof

AI claimed a solution to the 167-year-old Riemann hypothesis — and a human did, too
Image source: TechCrunch

A Historic Week for Mathematics and AI

It has proven to be a remarkable week for mathematician Levent Alpöge, according to reporting by TechCrunch. Alongside watching his employer, Anthropic, transition into the first publicly traded frontier artificial intelligence laboratory, Alpöge announced what appears to be a solution to one of mathematics' most celebrated and enduring challenges. The Riemann hypothesis, which dates back to 1859, stands as one of the official Millennium Prize math problems. At its core, the hypothesis conjectures the existence of a hidden underlying structure that governs the appearance of prime numbers. If the newly claimed result withstands total scrutiny, the mathematical community could be entering uncharted territory.

In an essay announcing his findings, Alpöge detailed the humorous and slightly chaotic timing of his breakthrough. He described messaging Anthropic CEO Dario Amodei late on a Sunday night to seek advice on how best to announce a Lean-compilable solution to the historical problem. The realization of this goal has left the mathematics community buzzing with excitement, confusion, and awe, with experts actively puzzling through the complex details in online discussion forums.

Inside the Technical Details and the Lean Verification

The mathematics community has taken to platforms like Zulip to dissect the intricacies of Alpöge's announcement. Experts such as Chris Birkbeck, a number theorist at the University of East Anglia, shared technical breakdowns of the work with colleagues. Birkbeck noted that the cohomology bound is seemingly proved via an automorphy-lifting argument. Furthermore, the argument utilizes the existence of Galois representations attached to regular algebraic, polarized cuspidal automorphic representations of GL_n over CM fields. It additionally incorporates base change, Jacquet–Shalika, Clozel’s purity, Bernstein–Zelevinsky, the main theorem of CM, Deligne–Ribet, and several other advanced mathematical frameworks.

While such high-level mathematics can quickly overwhelm observers, the reactions within the expert community indicate the profound significance of the work. Following Birkbeck's technical post, numerous readers reacted with astonishment. It will undoubtedly take considerable time for the broader mathematics community to fully comprehend the proof. However, experts express confidence in the formalized result due to its implementation in Lean.

Lean is an open-source coding language utilized to express mathematical concepts rigorously as code, enabling automated formal verification of proofs. Several prominent participants in the Zulip chat, including MIT mathematician Andrew Sutherland, have successfully confirmed that Alpöge’s solution compiles properly. Furthermore, updated reporting has verified that independent sources have successfully checked Alpöge’s Lean proof.

The Role of AI and the Quest for Human Understanding

Alpöge expressed a degree of personal guilt in his essay regarding releasing the proof without providing a readily human-intelligible explanation first. Nevertheless, driven by a childhood obsession with the Riemann hypothesis, he felt compelled to share the breakthrough immediately rather than wait. He characterized his journey of discovery as a stream-of-consciousness collaboration with Claude, describing how the AI model flew, danced, and sparkled near the finish line of the complex problem.

Despite the triumphs of the AI-assisted workflow, substantial work remains. For the achievement to carry full weight within the traditional mathematical canon, humans must eventually be able to understand the result and explain it clearly to one another. Alpöge has committed to pursuing this vital task. Additionally, Alpöge and his collaborator, NYU mathematician Tristan Buckmaster, released another important proof related to the Navier-Stokes equations, adding to the ongoing wave of mathematical breakthroughs involving frontier AI labs.

A Competing Human Solution Emerges

In a fascinating parallel development, the mathematical world received news of a completely human-generated solution to a related problem. Stanford professor and American Mathematics Association president Ravi Vakil shared a proof of the General Riemann hypothesis written by Stanford postdoc Alessandro Masullo. Vakil emphasized that the work is entirely human with no AI involvement.

According to Vakil, the original plan was to bring Masullo to Stanford for a year to polish the manuscript and announce it gradually after thorough peer review. However, given the rapid pace of current developments, releasing the current draft immediately was deemed the best course of action. Because this competing proof has not been expressed in Lean, experts note that it is still too early to confirm whether the proof checks out, let alone whether anyone outside the immediate circle fully understands it yet.

In a brief introduction to his proof, which represents the culmination of 12 years of dedicated work, Masullo offered a reassuring message to his peers. He wrote that if his work sends any message to mathematicians feeling discouraged by the rapid pace and tone of the modern climate, it is that researchers are still here, very much alive, and not going anywhere.

Source: TechCrunch

Post a Comment

To Top