@scientificamerican: A human-AI team may have solved the biggest problem in math. Yesterday Anthropic researcher Levent Alpöge posted a Claude-generated proof that he claimed solved the Riemann hypothesis. Alpöge’s proof is technically for the “generalized Riemann hypothesis,” or GRH, which is a bigger, broader and more powerful application of Riemann’s initial insight. Mathematicians are racing to independently verify Alpöge’s work, with preliminary and informal responses already emerging that the proof is indeed correct. But what would this actually mean for the rest of us? Would proof of the Riemann hypothesis make our computers faster, our online transactions more or less secure or our digital lives fundamentally different? The answers are murkier than you might expect. The most consequential impacts of Alpöge’s claim (if verified) wouldn’t really concern math at all. Mathematics underpins much of theoretical physics, engineering, computer science and other scientific disciplines, so an AI capable of proving the GRH today could, tomorrow, help researchers tackle problems in those fields that might otherwise be beyond reach. This would be a new way of doing science, and it signals a future, perhaps imminent, if not immediate, in which machines supersede the capacities of human beings for making fundamental intellectual discoveries. And that’s a sobering thought. Read more at the link in our bio. 🔗
Scientific American
Region: US
Saturday 21 November 2026 15:35:22 GMT
Music
Download
Comments
Brog :
maybe they should crack P vs NP next just so we know if we’re getting cooked in polynomial time
2026-11-21 16:42:40
9
MToasttoast :
I looked it up on AI and solving it is not of any real consequence
2026-11-21 21:51:25
0
drsleepofficial :
This is a big deal if it’s true
2026-11-21 21:47:11
0
Ryan Andrew K :
Levant Alpoge’s communication with Anthropic’s marketing and executive teams, lean team, work on finding kernel exploits need to be preserved for possible congressional and SEC hearings and investigations wrt possible securities fraud.
2026-11-21 21:05:29
2
leadkingapp :
Means that we are well cooked - SHOR will do its job soon RSA will be broken so will BTC
2026-11-21 16:10:45
2
Jeffery :
You talk about the good but what about the bad???
2026-11-22 00:43:21
0
Ryan Andrew K :
Except the frontier LLMs are extremely good at finding and things like lean kernel bugs and locally relentless at getting a “pass” via “gaming”. No independent and verified checker on a lean-runtime-independent has checked the proof yet. A separate matter is coincidence of flaws which these tools are also good at finding. I think many journalists visit zulipchat and lean communities to get a feel for what the “experts” are saying but this is actually a subset of believers not a truly decentralized and independent group. Then they come out of reading comments and professional analyses believing there is some sort of consensus that for GL (1) over Q that the RH and GRH has been proven with a 99 to 99.999% likelihood which is errant logic considering the selection effect. You as a lay influencer yapping about this miscalibrated sense of uncertainty and ambiguity only add to the noise— you’re a sucker. Also it’s a part of the RH.
2026-11-21 21:01:12
0
To see more videos from user @scientificamerican, please go to the Tikwm
homepage.