A Mathematician and Claude Say They've Cracked the Riemann Hypothesis. Now the Checking Begins
If you've ever heard that some math problems are worth a million dollars, this is one of them. The Riemann hypothesis has been around since 1859. It's about how prime numbers are spread out, and it sits on the Clay Mathematics Institute's list of seven Millennium Prize Problems, each with a $1 million reward. People have been trying to prove it for more than 160 years.
On Friday, Levent Alpöge, a mathematician at Anthropic, announced that he believes he and Claude have done it. And he says the proof covers the generalized version of the hypothesis too. According to IBM's Think news site, he didn't release it as a normal paper. He put out the argument in Lean, a proof assistant where a computer checks every logical step. He also admitted it's extremely long and hard to read.
That's what makes this so strange. Experts sound impressed, and they're also openly saying it'll take a long time to understand. James Maynard, an Oxford professor and Fields Medal winner, said that if it fully holds up, it's clearly a historic milestone, and that he suspects it'll turn out to be correct. He also said even checking the Lean verification will take time, and that really understanding why it works will take much more work. Youness Lamzouri of the University of Lorraine said he was in awe but thinks digesting the ideas could take months or even years. Independent mathematicians haven't verified it yet.
Terence Tao warned earlier this year that AI might produce answers faster than humans can understand them. This might be the first huge test of that idea. If a computer says a proof is correct, but no person can explain it, is the problem really solved?
What do you think? Does proof count if humans can't follow it yet?
AI may have helped solve math's most famous problem. Do we trust proof no human fully understands?
