Saturday, November 21, 2026

250,000 pages

That Riemann Hypothesis proof? The one done with AI by Levent Alpöge at Anthropic? It's over 250,000 pages long. Here's his four-page announcement.

Science:

Many mathematicians say they think Alpöge’s Riemann proof will hold up to scrutiny—and predict it will have far-reaching implications. “The blast radius extends to the whole of mathematics,” says Ghaith Hiary, a mathematician at the Ohio State University. “I didn’t think I’d see the day … but here we are.”

What’s far less clear is when—and if—mathematicians will manage to glean any understanding from the messy heap of code, which is larger than some entire computer operating systems. Researchers are left facing a bizarre reality: finally holding in their hands a long-awaited masterpiece that they can’t interpret without help from AI. The situation reinforces an existential lesson that mathematicians have drawn from AI’s descent on the field: “The point of math—and the point of science—is understanding why things happen,” not just collecting statements of truth, says mathematician Maksym Radziwill of New York University (NYU).

Mathematicians are starting to think it's right since it "compiles in Lean."

Friday, November 20, 2026

AI Scores Again - Riemann Hypothesis

A leading mathematician says he has, with the help of an AI, solved the Riemann Hypothesis, the most famous and important unsolved problem in mathematics. It's been unsolved for 167 years.


Where the Riemann zeta function is


If true, it would be the 2.5th Millenium Problem solved, all this year. (The one-half is because the mass gap of Yang-Mills theory hasn't been shown, though the existence of Yang-Mills theory is said to have been solved recently.)

ADDED: A second proof has been claimed, done months earlier without AI.

AI and the Near Future

The AI Futures Project, Nov 10th: 

METR recently estimated that Anthropic is now making AI progress about twice as fast as its historical rate (they would not be surprised by anything from 1.3x to 3x), and guessed that “perhaps ≈80%” of the current acceleration is due to AI. If they’re right, that implies an AI-driven acceleration that is 5 months ahead of the AI 2027 scenario.

This level of acceleration due to AI is very unexpected. Previously, Daniel, the most bullish of us, assigned a 50% chance of Superhuman AI Researcher (SAR)1 before 2029. SAR is roughly the point at which a company would rather fire all their human researchers and rely solely on automated AI researchers than have a team of the best human researchers in the world with no AIs.2

Now, all of us forecast that Superhuman AI Researcher (SAR) will likely arrive at some point in 2027 (assuming no substantial deliberate slowdown).3 We also predict a 35-65% chance that Top-Expert-Dominating AI (TED-AI), AIs that can automate all white-collar work, will arrive before 2028. While there were many factors and sources of evidence that influenced us, the single biggest source was the METR report.4.
80 precent!

I have to admit that this all seems almost impossible to me, even wildly preposterous. But that surely says more about me than about AI. Lots of people from Silicon Valley are talking about this on social media, and many see it a threat. Even the younger guys right in the middle of everything are sounding scary, and I've always thought that it's the younger guys in any corporation or industry who really know what's going on.
 
Here is Anthropic co‑founder Christopher Olah's full speech at the launch of Pope Leo XIV’s first encyclical, Magnifica Humanitas: On the Protection of Human Dignity in the Age of Artificial Intelligence, posted 5 months ago. I've started the clip where I think it's most important.

   

This does not look good for society. Even if it ultimately works out for humans, there is going to be a decade or two of turmoil, people losing their jobs, not knowing what they will do for money, for food, for shelter. I can't see that as being anything but very ugly and scary.

By the way, as of Thursday Forbes estimated Olah's net worth at $14.2 billion.

Monday, November 16, 2026

Podcast on AI taking over mathematics

Yesterday I listened to a podcast from "Radiolab" about the recent AI mathematics revolution you may have heard about. 

Radiolab is a science-oriented show on America's public radio, NPR. (I guess we have to call these "podcasts" now.) This podcast(!) was originally released about a month ago. I found it very interesting--I didn't realize mathematicians now consider an AI model to be a "world class mathematician better than any human mathematician." Experts label it an "emergent property" -- this ability emerged not by being designed, but by AI's reading all the mathematical literature out there done by humans.

The big shock came this past May after an OpenAI model solved the unit distance conjecture, disproving it. Then there was a Navier-Stokes fluid equation conjecture that it also disproved, finding an example where the equations "blew up" (gave a singularity).

Here's the timeline the guest gave for AI's development:  

  • 2024 - AIs were doing high school level mathematics
  • 2025 - AIs were as good as a very smart undergraduate studying mathematics
  • March 2026 - AI could do research level mathematics but was not the world's best mathematician
  • May 2026 - AI was solving problems that the best mathematicians had not been able to solve 

This is an incredible pace of improvement. In September OpenAI released a tranche of 700+ papers proving or disproving (they claim) a great many conjectures that had been made in the past.

Mathematicians are literally wondering if human mathematics is over.

Now ask, where will AI's math abilities be a mere year from now? 

Now imagine 10 years from now. Will AIs be writing about mathematics no human has ever imagined? That humans can't understand? That humans can't even make sense of? Will humans be able to understand its papers (which by the way have not submitted (I think) to peer review journals)?  

Scary. But also fascinating.

Sunday, November 15, 2026

El Niño Starts Pummeling the Planet

2026 will be the warmest year in GISS's records, which start in January 1880. 

It's only November and the temperature results for global temperature anomalies have come in for October:


Looks like all these records will be broken soon, if not already. 

For the globe, the annual temperature record will be the warmest year if (November+December) average is greater than 1.35°C. Looks like a done deal.

All four of these regions had the warmest October in their record. For the globe, the ranks of the last four months have been 1,1,2,1.

It's not even clear yet if the Nino3.4 region of the ocean has peaked in temperature:


(The line just below 2026's current position is 2015.) 

This is looking to be a very bad El Niño.

Wednesday, November 11, 2026

New Physics Book by a Friend: "The Great Atom Debate"

My friend Paul Halpern, a close buddy in graduate school, has a new book coming out, his 19th!

The Great Atom Debate
Ernst Mach, Ludwig Boltzmann, and the Battle for the Essence of Reality

"The little-known story of how two of history’s greatest physicists waged a war over the existence of the atom and the nature of reality itself."


"Nearly 2,500 years ago, the Greek philosopher Democritus argued presciently, but without evidence, that all matter is made of tiny indivisible entities called atoms. Other sages, similarly lacking proof, suggested that nature is continuous. By the later nineteenth century, though science progressed, the question of nature’s basic ingredients remained elusive."


"Two extraordinarily brilliant, incredibly stubborn Austrian physicists of that era—Ernst Mach and Ludwig Boltzmann—fought to determine whether flowing energy or discrete matter was the chief component of the universe. Mach treated atoms as useful fictions. Boltzmann, however, insisted atoms were a physical necessity. The titanic struggle rattled Boltzmann’s nerves, and he eventually descended into madness and suicide. Extraordinary experiments soon thereafter vindicated his lonely plea for atoms."


"Despite their differences, both thinkers influenced the young Albert Einstein, paving the way for relativity and quantum theory. Mach is best known today for his groundbreaking work on supersonic speed and shock waves."


"By connecting the clash of Boltzmann and Mach to today’s debates over what is fundamental, The Great Atom Debate does more than help us understand physics’ past. It helps us make sense of the unseen entities that loom over physics’ future."

(Amazon)

It’s being published on December 1st, so just in time for that peripatetic fat socialist to put in your budding little physicist’s stocking on the fireplace mantle. [Please solidly secure the stocking.]

The Great Atom Debate book cover

Saturday, November 07, 2026

Friday, November 06, 2026

"He Vanished on a Run Before We Could Marry"

I found this to be an amazing and moving essay. I thought the writing is sublime. Reminds me of Pam Houston (Cowboys Are My Weakness), if you know her writing.

Free link:
"He Vanished on a Run Before We Could Marry: If I am not a widow, what am I?" Beth Henshaw, NY Times, today

Thursday, November 05, 2026

Ever Upward

The ocean heat contents for both ocean regions are accelerating.

acceleration(0-700 m region) = 0.0089 ± 0.0003 (W/m2)/yr = 143 ± 5 exajoules/yr2

acceleration(0-2000 m region) = 0.0092 ± 0.0044 (W/m2)/yr = 149 ± 71 exajoules/yr2

uncertainties are 2-sigma.
Exa=1018.

Data are through the second quarter (2Q) of 2026. The uncertainties for the 0-2000 m region are much larger because there is less data. 0-2000 m data starts in 1Q2005 while the 0-700 m data starts in 1Q1955, so that's 86 data points vs 286 data points. Or maybe I made a mistake. Or both.

You can see the clear acceleration in the 0-700m graph and just about in the 0-2000m graph. Remember my acceleration calculations start at the beginning of each dataset, not at whatever happened around 2005.  



Monday, November 02, 2026

Binance Hack - 25% of NK's GDP

In last week's Binance cryptocurrency theft by North Korean hackers, $6.3 B was initially lost. According to the NY Times, over $2 B of that has been frozen by other crypto companies, halting the North Korean's attempt to money launder it.

North Korea's GDP is about $17 B/yr. The ~$4.3 B they retained is 25% of North Korea's GDP

Interesting Tidbit

Because the specific heat of ocean water is much larger than that of air, the top 2.5 m of the ocean holds as much heat as the entire atmosphere above it.


Source: Wikipedia, "Mixed Layers"