Claude Proves the Riemann Hypothesis
Kenneth Chang and Siobhan Roberts, reporting for The New York Times:
The Riemann hypothesis, often considered the most important unsolved problem in mathematics, may be the latest to be conquered by artificial intelligence.
On Friday, Levent Alpoge, a mathematician at Anthropic, posted a four-page note on his personal website saying that he had completed a proof of the hypothesis by collaborating with Claude, the company’s A.I. system. He did not include a traditional mathematical paper to describe the proof, but instead linked to computer code that he claimed would verify the result.
“Claude and I have had a great time together, with such breadth in recent months,” Dr. Alpoge said. “But this time, Claude far outpaced me near the finish line. No, Claude flew, and danced, and sparkled near that finish line.”
There are a lot of complicated feelings among mathematicians about this proof. Humans haven’t been able to solve it for 167 years, but a large language model solved it relatively quickly. The proof is staggeringly large, at about 250,000 pages, which no human could conceivably write and fact-check in any reasonable time.
My favorite opinion about artificial intelligence in math is from Dr. Terrance Tao, a mathematician at the University of California, Los Angeles, famed to be one of the greatest mathematicians of this generation:
I’ve never seen such brutal competition, especially with such money-backed resources behind it. We’ve learned almost nothing mathematically from what they did. Normally, when a human proves a great mathematical result, they can talk about what worked and what didn’t. But here, it’s like strip mining the landscape. You get all the fruits, but the landscape is barren afterwards. We need to develop new practices for doing mathematics responsibly in the AI era.
“We’ve learned almost nothing mathematically from what they did.” What a stunningly acerbic statement. I wish to posit a deep philosophical question that I have no intention of answering, but that I think is relevant to this latest news: What is the point of studying anything at all? Is the point of inventing novel mathematics merely to solve today’s problems quickly and easily and leave it at that? I would say no. What problems in Bernhard Riemann’s society did he think could be solved by proving this conjecture? Was there an immediate need in 1905 to prove relativity or discover novel radioactive isotopes in 1898? We learn and prove things — especially in mathematics and the natural sciences — not to solve immediate problems, but to enhance our understanding of natural phenomena. It is an inherently human endeavor to be curious.
Using AI to supplement this inherent human curiosity and help us answer our most complicated questions is humanity at its best. We have invented mathematical algorithms that can speak to us in our own languages to help us learn and explore natural phenomena. That might be a contender for mankind’s greatest achievement. But using AI to supplant human curiosity and creativity would equally be the biggest own-goal in mankind’s history. The purpose of studying mathematics, computer science, the natural sciences — really any academic discipline — is to better understand the world we live in and the people who we live with. If AI is replacing that curiosity and opportunity to learn about ourselves, then it is replacing a core tenet of humanity. If we are not learning anything from these AI-assisted discoveries, then there is no point to those discoveries to begin with.
So, no, one should not abandon their love and study of mathematics or any other science because AI is proving the oldest conjectures. We ought to learn from those proofs instead. And the people who think mathematics — similar to computer science — is “solved” because a computer can aid us in our inherently human quest to know everything deserve no place in academia at all. The dedication of one’s life to the study of a particular subject does not end by solving any one given problem. If that were the case, we should’ve stopped long ago. We should’ve ended our study of physics after Albert Einstein posited general relativity. That is the greatest physics problem ever solved, right? But we never stopped studying physics because one problem was solved — and similarly, we should never stop studying any field because our computers can help us study so much more than we ever could have alone.