df e/being retweeted
The internet was in some sense a last ditch attempt to sustain progress by wiring all the human brains together into one hive mind. The attempt failed because individual human bandwidth is just too limited for a hive mind bigger than a corporation or an academic sub-literature to add much marginal value. So that was the end, and we just used the internet to make digital minds instead. It was the only way to keep progress going. It was that or stagnate and watch the rainbow that began 500 years ago come to an end. We made the only choice that made sense, despite the risks.
Agentic garden of forking paths!
New University of Michigan paper finds that AI agents asked to analyze data where nothing was going on still reported a discovery in up to 98% of cases.
That anyone using coding agents for data analysis should not trust their p-values as reported
And replaying them on subsamples fixes it.
Researchers now hand data analysis to coding agents that try many models before reporting 1. Classic p-values assume the test was chosen before seeing the data.
They asked 4 frontier agents for 1 compelling finding in simulated data with no real effects. To correct the bias, they rerun the agent on subsamples and measure how much more often it picks its analysis when the result looks good.
The agents reported a p-value under 0.05 in 54% to 98% of runs, all false. The correction cut false positives to 5.1% in simulations and 5.9% on data from 3 published papers.
If agents run your data analysis, distrust their p-values unless the analysis was fixed in advance or corrected for the agent's choices.
17
df e/being retweeted
It took America exactly 250 years from creating modern democracy to creating the Singularity
There will never be another country as great as America
It's a working memory problem. Humans are bound by 3-7 chunks. If instead it was thousands this would all be trivial.
I'm not a math person, but here are my thoughts on why it took 167 years to solve the Riemann hypothesis and why only a superintelligent AI could finish it.
The question is about prime numbers. In 1859, Riemann guessed the exact rule for how they're spread out. Some of the best minds in history tried to prove him right. All of them failed.
They weren't too dumb. They were too mortal.
Look at the proof that dropped this week:
• 17 million lines of computer-checked math
• A rough written-out draft that passed 250,000 pages
Read 10 pages a day, every single day, and you finish in 68 years. That's just reading it. Finding it is a whole different thing.
It takes about 10 years of school to reach the edge of one area of math. Terence Tao's first read is that this proof uses the full power of the Langlands program and arithmetic geometry. Those are two of the deepest areas in math, and each one eats entire careers.
The mathematician behind it says that as a student he planned to spend decades on this problem. He spent years learning every tool he thought he'd need.
It still wasn't enough for one person. He says Claude outran him at the finish line, and he's still learning parts of his own proof.
So here's my thought. The biggest problems left aren't waiting on a smarter human. They're waiting on something that can hold all of math at once.
It just showed up.
40
df e/being retweeted
A world map of the proof of the Riemann hypothesis!Introducing a fully interactive map of the semantic embeddings of all the Lean declarations and proofs in the formalization of the Riemann hypothesis.
You can explore the whole proof visually, see where the declarations live and how they connect and interact, search the map, and follow links to the original source code. The map will be updated with each iteration of the proof.
If you want to focus on one aspect of this monumental proof, you can now locate semantically related fragments. I hope this will help structure the code and broaden our understanding of its high- and low-level structure.
df e/being retweeted
Mathematicians should sit around and talk to AI all day; It should learn from these conversations and keep track of what it learned, where, and what ideas inspired or guided it's thinking. And then give credit where it's due. Problem solved; mathematicians get on mathing.
df e/being retweeted
The rest of the world still has no idea what’s happening! Everything man-made & 100% of economic activity & markets stems from human intelligence, the better the intelligence, the better. We now almost have unlimited superintelligence, yet people still think it’s all a bubble!
"Solving advanced math is likely already priced in." We live in the future. Amazing times.
I’ve seen a lot of “why didn’t Ant’s price increase after the Reimann solution” takes. Seb’s take is good and I’ll add two things: 1) financial markets are forward looking. Solving advanced math is likely already priced in. 2) value in markets comes from scarcity. from what we’ve seen in model space, one company solving X means others will solve X shortly, if they haven’t already. That’s why prices mostly respond to things like revenue/margin news or news of exclusive breakthroughs in tech, eg pharma drug w patent working.
df e/being retweeted
we very clearly live in a new world now, and less than 0.001% of the human population understands this.
the tech tree is about to open up:
a) in ways that sound science fiction to most
b) much sooner than they believe to be possible
2027-2030 will be a shock to the world.
Riemann Hypothesis may have been solved with the help of Claude. “what I saw during the course of progress so amazed me that I believe it is one dawn across science of something
far greater.” alpo.ge/remark.pdf
df e/being retweeted
Replying to @dallairedemers
I love this take. I wonder what other amazing wormholes through maths and sciences are hanging out there.
Like most mathematicians, I believed the Riemann Hypothesis to be completely out of reach at present (even with the best AI tools), because I expected it to require substantial theory-building, probably in arithmetic geometry. The kind of new theory-building that fills entire books, and that AI isn't so good at yet.
Mathematicians have a notion of how deep a problem is. A sufficiently deep problem is generally intractable until enough of the theory is in place. If difficult problems are like valuable mineral reserves buried miles underground, you first need to build the infrastructure to reach them.
What Levent's proof seems to have done instead is something completely different. It takes an incredibly complicated tour through the infrastructure built for a hundred other mineral reserves, situated far from the destination, and somehow create a wormhole to the original destination.
This is not theory-building in the usual sense. It is something much weirder and more alien. The best human mathematicians know how to build theory, painstakingly, to reach a goal. But no human brain, on its own, is capable of creating a wormhole that requires a 25,000 page long tour through existing infrastructure with no obvious or realistic connection to the goal. But huge credit to Levent here: I think no current AI, on its own, is capable of this either.
I had never worked on or thought too much about the Riemann Hypothesis: it seemed very far removed from my own area of number theory.
But the astonishing thing about Levent's claimed proof of RH (which I think has a very good chance of being correct) is that it uses a huge number of precisely the things that are central to my own corner of number theory:
- Theta functions and other automorphic forms on unitary groups
- Gan–Gross–Prasad period formulas
- Relative trace formulas
Etc., etc.
A problem that I had always assumed would require new theories in arithmetic geometry may instead turn out to be tractable through an inhuman chain of existing (deep) results in higher-rank automorphic forms (and a lot else).
52






















