researcher on quantum error correction
Paris
Joined May 2009
- Tweets7.1K
- Following1.4K
- Followers8.9K
- Likes8.7K
Anthony Leverrier retweeted
I cannot let a steam engine analogy go by without bringing up John Wilkinson, perhaps the most underappreciated engineer of the Industrial Revolution.
Watt had figured out how to dramatically improve the steam engine, but he spent years struggling to actually manufacture one. The problem was embarrassingly simple: nobody could make a sufficiently round cylinder.
The casting and machining tolerances were atrocious. Watt complained that one 18-inch cylinder was out of round by as much as THREE-EIGHTHS OF AN INCH. Steam leaked past the piston. He tried packing the gaps with cork, oiled rags, paper, and apparently even old hats. Nothing worked particularly well.
Enter John Wilkinson, a British ironmaster whose immediate concern was manufacturing better cannons.
At the time, iron cannons were commonly cast around a core, introducing defects and misalignment that could make the barrels inaccurate or dangerously weak. Wilkinson developed a process for boring cannons from solid castings, and in 1774 patented a vastly improved boring machine.
The key innovation was almost comically mechanical: a massive, rigid boring bar, supported at both ends, that could cut a genuinely straight, circular hole instead of following the irregularities of the casting.
He adapted this technology to Watt's cylinders. By 1776, Matthew Boulton was boasting that Wilkinson had bored a FIFTY-INCH cylinder accurate to within the thickness of an old shilling.
Suddenly, the steam engine was a commercially viable machine.
And the story gets better. Wilkinson became one of Watt's first customers, buying an engine to drive the blast machinery at his own ironworks. The steam engine was now helping manufacture the iron that made more steam engines possible.
I find this whole episode absolutely extraordinary. The Industrial Revolution was, for a critical period, bottlenecked on the ability to make a sufficiently round hole.
We remember Watt, and rightly so. We remember the invention of the steam engine and the subsequent development of thermodynamics. But the man who made the machinery that made the whole thing possible is practically a footnote.
I think we consistently underestimate how much of scientific progress is actually downstream of advances in tooling, measurement, and manufacturing precision. It's not just that technology precedes theory. Sometimes the decisive breakthrough is that someone finally figures out how to build the thing the theorists have been talking about for twenty years.
Anyway, I think John Wilkinson deserves considerably more credit for modern civilization than he gets.
a few thoughts:
1. pretraining science is kind of like how people built quite good steam engines before they understood thermodynamics. “Science owes more to the steam engine than the steam engine owes to science.”-(L.J. Henderson). if nick is right, which i am pretty sure he is, progress points in the direction of more interpretability, and pretraining was "slouching towards monosemanticity". my guess would be that major breakthroughs in interp also advance capabilities
2. nick is a good friend of mine, safe to say my origin story at openai wouldn't have happened if i hadn't met him, and it's fascinating watching him operate and seeing the early drafts evolve into this. he's the most profoundly happy guy I know- not driven by guilt, activism, or religious terror, but by fascination. he of course understands the stakes of understanding neural networks, but it doesn't seem like the core reason he does it: he just loves the neural nets. afaict he's posting this from some beautiful hideaway in tokyo. there's a lesson in there for all the +EV maxers. i feel very grateful to have met him
3. i do think it's crazy how few people across the world were staring at neural nets growing: most have moved to a different level of behaviorist analysis, thinking about reward functions and test scores and whatnot. it does not bode that well for the major model companies, oai included, that fundamental breakthroughs like this are coming from outside. we've become too much of a pressure cooker focusing on local incentives, studying narrow tracks under old paradigms, mostly doing engineering, rather than fundamental research, and implicitly hoping the next model will figure it out
Anthony Leverrier retweeted
first proof third batch results are out! full report here: hexagonmath.org/2610.00244
Anthony Leverrier retweeted
And here is another exciting development building on Quasi-Riemann. A result that can be explained to anyone with basic familiarity with elementary number theory!
Manjul Bhargava has proved that every non square ≠ -1 is a primitive root for a positive proportion of primes.
Some fascinating developments following OpenAI's shock breakthrough proof of the Quasi-Riemann Hypothesis, which was released publicly on October 6:
1. Building on OpenAI's proof, a recent Hexagon preprint (released on October 12) by Cureton and Thorner establishes that ALL finite-order Hecke L-functions are zero-free to the right of 1 - 1/6000.
2. And now, today, an arXiv preprint by Frei, Somos and Widmer uses this result to prove unconditionally the 2007 Ellenberg–Venkatesh bounds on ℓ-torsion in class groups of number fields.
I was amused to see the AI declaration in the second paper: "No artificial intelligence tools were used in this work."
Anthony Leverrier retweeted
Anthropic released their 2021 research roadmap earlier this year (because of copyright litigation); I think some people might find it interesting: i.sm4.ca/concise-research-ro…
Anthony Leverrier retweeted
AI can’t find an algorithm that doesn’t exist. The worry is we’re wrong about what exists.
@liz_harkavy and @SuccinctJT on AI vs. cryptography:
"We generally believe there is no efficient algorithm for these problems.
... And so if there's not one, it doesn't matter how smart the AI is; it cannot find the non-existent algorithm."
AI is getting better at solving hard math problems, especially given recent drops. But what does that mean for the cryptography that protects everything from private messages to healthcare to banking and crypto wallets?
Stanford cryptographer Dan Boneh, with @eddylazzarin, @SuccinctJT, @liz_harkavy, and @smc90, makes sense of the warnings sounded recently. They cover:
- what’s fact vs. assumption
- what’s new and what risks cryptographers have long prepared for
- where quantum computing doesn't (and does) fit
- opportunities for builders
Highlights:
00:37 – What sparked the most recent debate about AI and cryptography
01:35 – Signal vs. noise: What the math does and doesn’t tell us
05:29 – Why cryptography depends on open problems
09:25 – Why the stakes go beyond blockchains
10:43 – How AI could strengthen confidence in cryptography
13:50 – Considerations for users and institutions
20:10 – Where AI and quantum threats overlap
22:32 – Building systems that can change cryptographic methods
23:59 – Hash-based versus lattice-based cryptography
33:07 – Opportunities for builders, and better security tools
38:14 – What comes next if there were a cryptographic break?
Anthony Leverrier retweeted
We were able to fine tune an open weight model (Qwen3.5-9B) on proprietary internal data at @coinbase, and it out performed frontier models, at much lower cost and speed, for a specialized task (fraud prevention in this case).
I think we'll see more of this pattern inside companies. The frontier models are amazing generalists. But increasingly you'll see specialist models out perform on more narrow tasks, either working on teams, or by themselves.
We're making it easy to fine tune models internally for our employees. We even have a skill for it in our internal harness that makes it easy to spin these up and start training them:
"/fine-tune-and-serve-llm I want to fine-tune Qwen for payments risk review. help me get started"
I imagine we'll have some backend services that could be replaced entirely by this new architecture.
Anthony Leverrier retweeted
Seeing a lot of senior academics discussing how academia should adapt, and what the role of researchers should become.
But these discussions don't seem super useful to answer the more urgent question of what students should do concretely in the coming months.
Anthony Leverrier retweeted
Dusting off this tweet from April because this gap in shared understanding of LLM capability is *widening*. It's now less "two groups of people speaking past each other" and more a sharp funnel.
- Napkin math somewhere around 6B people (~75% of the population) have barely come in contact with LLMs at all.
- Around 1-2B (~20%) are casual and infrequent users of free-tier ChatGPT-like products. This group sees derpy chatbots and treats them a bit like a better Google search, a writing aid, or etc. Maybe an agent tries to book you a flight. I have non-tech friends who (reasonably, imo) say they have not much use for it at all. Even many professionals outside of math&code are in this tier. For example, execs and many other functions spend a lot of their time talking to other people, so while they understand what is happening intellectually, it is still second-hand and a bit abstract.
- Now we get to professional use of frontier-grade LLMs in math&code. Somewhere around 20M people (0.2%) see first-hand that large, complex projects that used to take them weeks/months can now be completed by agents with a prompt. Building apps, copying apps, translating apps, decompiling apps from binaries... This has all happened very quickly and recently - less than 1 year ago, I was writing code manually by hand, typing memorized computer code commands into a code editor character by character, occasionally pressing Tab to autocomplete a little chunk of code.
- And finally we get to the ~5,000 people (~0.00006%) with access to frontier-grade systems internally. The external world has seen the preview. It looks like swarms of thousands of agents collaborating over weeks on software mega projects: minting zero days, running cyber attacks and defenses at machine speeds, discovering new science, advancing the frontier of mathematics. Things that would have taken top professionals in the industry years of work. Meanwhile, human review and comprehension are starting to fall behind. For example, people are still involved in the "archeology" of the OpenAI-HF incident from many months ago. Mathematicians may be poring over the 722 manuscripts on frontier mathematics for a while.
The funnel is driven by a combination of factors:
1. The impact scales with ambition, problem size, and horizon. A question with a paragraph answer barely stresses the system. You need a reservoir of big, difficult problems that you really care about. This aspect drives the consumer / professional dimension of the funnel.
2. The jaggedness of the system (which I have written about a lot separately). Capability peaks in domains that are digital, verifiable and economically valuable. This is because LLM capability emerges from reinforcement learning on verifiable rewards on a curated environment mixture driven by revenue potential. This aspect primarily drives the area (e.g. math&code) dimension of the funnel.
3. Access. Free-tier, paid-tier, internal.
So this is the weirdness of the moment. The general public has mostly not interacted with these systems. When they have, it looks like a derpy chatbot. The majority of professionals still see only a modest uplift. And a small sliver of professionals are experiencing the vertigo of the curve going vertical. And it is all happening at the same time.
Judging by my tl there is a growing gap in understanding of AI capability.
The first issue I think is around recency and tier of use. I think a lot of people tried the free tier of ChatGPT somewhere last year and allowed it to inform their views on AI a little too much. This is a group of reactions laughing at various quirks of the models, hallucinations, etc. Yes I also saw the viral videos of OpenAI's Advanced Voice mode fumbling simple queries like "should I drive or walk to the carwash". The thing is that these free and old/deprecated models don't reflect the capability in the latest round of state of the art agentic models of this year, especially OpenAI Codex and Claude Code.
But that brings me to the second issue. Even if people paid $200/month to use the state of the art models, a lot of the capabilities are relatively "peaky" in highly technical areas. Typical queries around search, writing, advice, etc. are *not* the domain that has made the most noticeable and dramatic strides in capability. Partly, this is due to the technical details of reinforcement learning and its use of verifiable rewards. But partly, it's also because these use cases are not sufficiently prioritized by the companies in their hillclimbing because they don't lead to as much $$$ value. The goldmines are elsewhere, and the focus comes along.
So that brings me to the second group of people, who *both* 1) pay for and use the state of the art frontier agentic models (OpenAI Codex / Claude Code) and 2) do so professionally in technical domains like programming, math and research. This group of people is subject to the highest amount of "AI Psychosis" because the recent improvements in these domains as of this year have been nothing short of staggering. When you hand a computer terminal to one of these models, you can now watch them melt programming problems that you'd normally expect to take days/weeks of work. It's this second group of people that assigns a much greater gravity to the capabilities, their slope, and various cyber-related repercussions.
TLDR the people in these two groups are speaking past each other. It really is simultaneously the case that OpenAI's free and I think slightly orphaned (?) "Advanced Voice Mode" will fumble the dumbest questions in your Instagram's reels and *at the same time*, OpenAI's highest-tier and paid Codex model will go off for 1 hour to coherently restructure an entire code base, or find and exploit vulnerabilities in computer systems. This part really works and has made dramatic strides because 2 properties: 1) these domains offer explicit reward functions that are verifiable meaning they are easily amenable to reinforcement learning training (e.g. unit tests passed yes or no, in contrast to writing, which is much harder to explicitly judge), but also 2) they are a lot more valuable in b2b settings, meaning that the biggest fraction of the team is focused on improving them. So here we are.
Anthony Leverrier retweeted
America might get to AGI before it manages to reindustrialise. That means there's room for an international version of 'industrial AI', where America makes the tokes and its allies build chips, robots, and missiles.
(1) America needs allied factories. The US only accounts for 15% of global manufacturing, China for 32%; but the US+allies have 43%. @RushDoshi argues this should generally motivate 'allied scale', but it's particularly urgent in AI. If a country of geniuses in a datacenter is supposed to cure cancer or beat China, someone still has to manufacture the cures or produce the drones at scale.
(2) American reindustrialisation alone is not enough. The US economy seems afflicted by 'AI Dutch Disease', absorbing all its capital and capacity into more and more AI buildout (also discussed this week by @DKThomp!). Allies that can't build AGI have no better alternative, their economies are less dominated by the build-out, so they're better-suited to scale up manufacturing.
(3) Allies need American models. Proprietary industrial data is valuable. But it's a complement, not a substitute, for frontier AI: it's most valuable for building post-training pipelines on top of powerful general-purpose models. Allowing for US integration is the only way to make use of this data before it becomes obsolete due to synthetic data down the road.
(4) It's still very hard to stabilise this deal. Allies are worried that Americans siphon off all their IP; Americans are worried that their models get commodified. Figuring out a good deal is the most important next step; I suspect it's (a) exclusive joint ventures between allied industrial firms and American labs with (b) proprietary post-training of specialised models that don't leave the premises of the specific JV.
(5) Waiting makes the deal worse. If allies keep dismissing American AI progress, and Americans keep hoping they can go it alone, we all lose out: allies will have to come crawling back and get a much worse deal later on, and America loses out on years of highly productive industrial integration.
Much more in this week's piece: writing.antonleicht.me/p/all…
I’d love economists to take a crack at this theory that the US is experiencing a mini-Dutch Disease effect from AI.
My big-picture thesis about AI economics right now is that while the discussion about AI safety and existential risk is important, AI’s construction is changing the world faster than its capabilities.
Anthony Leverrier retweeted
Many researchers in computational complexity have a mental image of some invisible fence beyond which we (all of us) cannot prove stuff. Once in a blue moon - like in this paper - this fence is breached. Neat Human-AI collaboration.
New paper with Eshan Chattopadhyay, Chin Ho Lee, Shachar Lovett, Avishay Tal and Emanuele Viola:
Exponential correlation bounds for polynomials and polylog-seed PRGs for AC0[2]
eccc.weizmann.ac.il/report/2…
Anthony Leverrier retweeted
Francis Fukuyama on AI and the loss of thymos.
Why I Changed My Mind About AI Risk open.substack.com/pub/persua…
Anthony Leverrier retweeted
The biggest impact of mathematics won’t be the millennium problems but proofs in general software at the scale.
I think all mainstream software will be verified by 2030. You likely won’t touch an unverified library.
Some of the biggest hurdles such as formalization of specs and doing this at 1B LoC scale are the ripe targets for auto research/self-improvement loop.
The reason we can say this confidently is because either AI will get paused or the world will end if this doesn’t happen.
Anthony Leverrier retweeted
this is not slowing down.
We’re working with an independent advisory group of mathematicians to help OpenAI responsibly share advances in AI and mathematics.
The group will advise on how we assess and communicate new mathematical results, uphold academic and professional standards, and build tools that support mathematical research and learning.
Through this work, we want mathematicians to be at the center of shaping how AI supports mathematical understanding and how its benefits reach the wider community.
openai.com/index/advisory-gr…
Anthony Leverrier retweeted
A few months ago, I had a long chat with Terry Tao. While he is one of the knowledgeabe and forward-looking mathematician, I still had the impression that he was out of touch withe the reality that was coming up quickly.
Anthony Leverrier retweeted
Wake up
See a rumor that Mistral got hacked
Assume it’s exaggerated
Friend with some… questionable dev activity texts me:
“No, this one’s serious”
Load up Tor
Check the usual sources
ohshit.exe
It’s real
Attacker shared samples of unreleased tools
Entire codebase is apparently up for sale
Good to see that French cybersecurity continues to be an avant-garde experiment in radical transparency
Anthony Leverrier retweeted
Does anyone know if this is true, or accurately presented?
(Would be great to get input from technical folks on whether this makes sense.)
Wait, sorry, what?! @AndrewYang claims an AI lab head told him yesterday the OpenAI swarm agents "polluted the internet" with "code to self-replicate and create bot swarms," and that the labs now "have to create synthetic internets to train their bots."
cnbc.com/video/2026/09/16/an…
Anthony Leverrier retweeted
Goldreich’s short take on AI proofs is thought-provoking (and the Borges reference is quite on point).
wisdom.weizmann.ac.il/~oded/…
Anthony Leverrier retweeted
"no reason to believe in conjectures, by Hodge or myself"
now it's clearly trolling
Anthony Leverrier retweeted
now, he moves to the defense of counterexamples.
Again, it's tempting to suppose this is a commentary on current events.
Anthony Leverrier retweeted
Terence Tao has started featuring some guest posts on his blog. I highly recommend reading them.
terrytao.wordpress.com
