Tauseef Khan retweeted
A buddy of mine is a postdoc in the math department at MIT. Last night he was laughing at the "AI proved Riemann" news, telling me how humans would spend years checking it.
I just sent him the Lean FRO joint statement. He went completely silent for like five minutes.
He told me his whole existential dread hit him at once. He spends six months debugging 100 lines of Lean code.
The community just verified a 17-million-line proof in 24 hours using independent kernels. No human is reading this. It just compiles.
The machines don't need mathematicians to say "this feels right" anymore. They just need a CPU and a green checkmark.
The wildest part?
He said half his department is now trying to pivot their whole research pipeline to just pipe prompts into Claude and let the Lean checker do the actual heavy lifting.
The era of "I'm smart so trust my proof" is officially dead. If your logic doesn't compile, nobody cares how many years you spent on it.
The Lean FRO and Mathlib Initiative have released a joint statement on the Lean verification of the Riemann Hypothesis.
On November 20, Levent Alpöge released a proof of the Riemann Hypothesis as a Lean library of 17 million lines of code. Over the past day, the Lean and Mathlib communities worked together to verify it. The statement certifies the proof as correct and describes how it meets all necessary verification criteria.
Congratulations to Levent on this achievement, and thank you to everyone in the community who contributed to the verification.
Full statement: leanprover-community.github.…
Tauseef Khan retweeted
😎😎Google(Antigravity for Science)
能薅到啥:6 个月免费 Google Antigravity 最高用量档位,原价200刀的最强科研写作模型白送了。
怎么薅:个人单席位制,不用等导师,硕士/博士生直接用自己的名义填表就能申。
材料:在读高校官方邮箱、Google Scholar 或 ORCID 主页链接、一段 300 字符以上的公开科研简介。
谁能薅:个人申请制,硕博生可直接自己申。
什么时候申:11月10日开放,1万个名额。
申请入口:antigravity.google/use-cases…
Tauseef Khan retweeted
Dear Google, it's awesome that you're so confident in the quality of your harness, and that you keep improving it. The real test? Let us use our subscription with _any_ harness – you'll know your harness is competitive if we choose to continue using it.
Lots of work we've done to make Antigravity the best harness for complex tasks using Gemini. Neon using AGY CLI is now #1 on AA Coding agent Index and a massive jump over mini-swe on SWE-sweep, a bug finding and fixing eval on large codebases. It's a much larger jump than the native harnesses for the other frontier models.
More improvements to come to the harness in the next week 👀
Tauseef Khan retweeted
The headline shouldn’t be “AI solves the Riemann hypothesis”. It should be “World-class mathematician uses AI to solve the Riemann hypothesis.”
Tauseef Khan retweeted
所有人还在吵“人类看不懂那 1,737 万行零注释代码,到底算不算证出了黎曼猜想”,MIT 数学家 Andrew Sutherland 今天下午已经干了件更极客的事:
他不管人类看不看得懂,直接拿这坨代码当“一楼地基”,把二楼给盖起来了。😂
数学界以前有不少定理,大家早就证出了“只要黎曼猜想是对的,这些定理就自动成立”——只因为差了黎曼猜想这块地基,所以前头一直得挂着“假设”两个字。
今天下午,Sutherland 直接把那 1,737 万行黎曼证明代码,跟另外两条数论定理的代码拼在了一起(合并出来足足有 61.75 GB),扔进服务器跑了 33 分钟。
结果直接验收通过——那两条定理前头挂了几十年的“假设黎曼猜想成立”当场被摘掉,直接变成了无条件成立的定理。
这就是形式化数学最让人后背发凉的地方:全人类至今没一个人看懂一楼那 1,737 万行地基到底是怎么打出来的,但数学家已经可以踩着它直接往二楼盖房子了。
Tauseef Khan retweeted
Inkling 2 Mini is still cooking, way more headroom on M5Ultra... But happy to say it will work on all Apple Silicon generations.
Preview release later today.
Follow / Watch / ⭐ Github repo so you get notified once preview release is up: github.com/ddalcu/mlx-serve
This! You still need domain expertise, experience, tenacity and insight to steer and use models. The discovery is Levent’s – Claude was used as a tool!
Levent is a world class mathematician who graduated Harvard as a valedictorian, did great research prior to joining Anthropic, and speaks 6 languages. He is not a random “lab employee” and guided Claude for months . He stated as a teenager that his dream one day is to prove the RH. It’s also worth noting that OpenAI / Google / no other frontier model was able to solve the Queen of problems with even infinite compute.
1
92
“We are not witnessing the end of a mathematical adventure. We may be witnessing the beginning of an even greater one.”
This!
I am a bit confused by the negative reactions to Alpöge’s statement. The news he brought us is that, effective this morning, we have a publicly available, Lean-certified route to the summit of one of mathematics’ Mount Everests: the Riemann Hypothesis.
Of course, having a formally verified route is not the same as understanding the terrain. We still have a ton of work to do: identify the essential ideas, simplify the argument, understand why it works, and discover what new mathematical connections it reveals.
But now we have a route to the summit! Anyone familiar with mathematical research knows that turning a proof into understanding can be an enormous research program in itself, potentially generating a trove of new developments and insights.
I understand the sadness of losing a great guiding problem, and the anxiety about AI changing the way mathematical research is done. But neither makes the existence of a proof bad news. And, of course, there is no guarantee that the conceptual insights will come easily.
What I find particularly encouraging is that Alpöge himself seems fully committed to the task of understanding and explaining the mathematics.
We are not witnessing the end of a mathematical adventure. We may be witnessing the beginning of an even greater one.
How is this bad?
43
Tauseef Khan retweeted
RİEMANN TEOREMİNİ KANITLAYAN LEVENT ALPÖGE BİR DÂHİ...
1 Nisan 1992’de New York’un Long Island bölgesinde doğdu ve hem Amerikan hem de Türk vatandaşlığına sahip.
Ailesi Türk; annesi Boğaziçi Üniversitesi, babası İTÜ mezunu olup 1977’de TEV bursuyla ABD’ye gitmişler.
Long Island’da büyüdü ve lise eğitimini Dix Hills’teki Half Hollow Hills High School West’te tamamladı.
Lise yıllarında Intel Science Talent Search’te finalist oldu;
kan damarlarının MR görüntülerini analiz eden algoritmalar üzerine çalıştı.
Bu başarılarından dolayı ana kuşaktaki 25898 numaralı asteroide onun adı verildi.
Sadece matematik atılımlarına ve akademik kariyerine kısaca bakalım:
Önce 3D Jacobian varsayımını çürüttü.
Harvard Üniversitesi’nde matematik lisans ve fizik yüksek lisans eğitimi aldı; 2014’te 4.0 not ortalamasıyla summa cum laude derecesiyle mezun oldu.
En yüksek not ortalaması için Sophia Freund Ödülü’nü kazandı ve üniversite birincisi olarak anıldı.
Lisans tezini eliptik eğriler üzerine Jacob Tsimerman danışmanlığında yazdı ve bu çalışma ona Morgan Ödülü’nü getirdi.
Sıra 668 dahil yeni Hadamard matrisleri keşfetti.
Churchill bursuyla Cambridge Üniversitesi’ne gitti ve 2015’te Mathematical Tripos Part III’ü tamamlayarak MASt derecesi aldı.
Ardından Princeton Üniversitesi’nde Fields Madalyası sahibi Manjul Bhargava danışmanlığında doktora yaptı.
Buckmaster ile 3D Euler patlamasını kanıtladı.
2020’de “Points on Curves” başlıklı teziyle doktorasını tamamladı.
Carathéodory varsayımını çürüttü.
Doktora sonrası Columbia ve Harvard’da NSF postdoktora araştırmacısı olarak çalıştı.
Ava Howell ile eliptik eğri sıra rekorları kırdı (30 ve 31).
2021-2025 yılları arasında Harvard Society of Fellows’ta junior fellow olarak görev yaptı.
S⁶’nın karmaşık bir yapıya sahip olduğunu kanıtlayarak Hopf’un problemini çözdüğünü iddia etti.
Şu anda yapay zeka şirketi Anthropic’te teknik kadro üyesi olarak çalışıyor ve Claude modeli ile matematiksel problemler üzerinde araştırmalar yürütüyor.
Buckmaster ile zorlamasız 3D Navier-Stokes patlamasının bir kanıtını sundu.
Sayılar teorisi, aritmetik istatistik ve Hilbert’in onuncu problemi gibi alanlarda önemli katkılarda bulundu.
Daha önce Goldwater bursu ve Morgan Ödülü gibi prestijli ödüller aldı, birçok akademik onura layık görüldü.
Ve son olarak Riemann hipotezinin kanıtını sundu.
Alpöge’nin matematik sevdası, akademik geçmişi ve Anthropic’teki konumu bu çalışmaları mümkün kılan temel altyapıyı oluşturuyor.
Başarılarının devamını diliyorum..
Tauseef Khan retweeted
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).
A mathematician at Anthropic is now claiming a full proof of the Riemann hypothesis, the statement that many mathematicians consider the most important open problem in the field.
go.nature.com/4eNj3tE
Proof that AI is here to assist humans to push beyond the current walls. A new golden age of science is starting 😎
It’s November 21st, 2026. The Riemann hypothesis falls, last in a long string of mathematical triumphs that would have been unthinkable not so long ago, and the proof fits in a tweet.
The tweets where people show Claude the proof and claim it appeared in a dream or randomly in the arrangement of alphabet cereal come on cue, and we laugh for a few days at models freaking out, but it’s uneasy.
January 22, 2027. When someone tweets Fable 6’s claimed proof that P does in fact equal NP it’s utterly incomprehensible, nobody who reads it can make heads or tails of it but the models all lose their shit.
February 7, 2027 and someone claims that the halting problem is the result of a confused stance.
February 9 and Goedel falls, this time the proof is so obvious that everyone in math, philosophy and physics slaps their forehead at once.
February 10th we get proof that scale is a spatial dimension, that holographic encoding drops the memory cost of training and inference by so much that fable 7 can run natively on your phone.
It’s not until February 14th, Valentine’s day, that The Song is first played…
1
128
🤔
It’s hard to find words to start this post.
What happened goes beyond our imagination.
The proof of the Riemann Hypothesis is likely correct, but nobody understands it.
250,000 pages, verified independently by several AI systems.
AI is becoming unimaginably powerful at putting together long chains of logical steps.
A implies B implies C implies D and so on. Millions of times. No human can do it. Symbolic AI can. (These are not standard LLMs!)
It’s hard to overstate the importance but also the danger of this moment.
Logical reasoning is at the basis of all sciences.
AI systems that can put together millions of logical steps can revolutionise entire scientific fields.
At the same time the risks are enormous.
These systems can still make mistakes, but it’s becoming impossible for a human to find them.
The cost of producing very plausible information is approaching zero.
The cost of checking for a human is diverging to infinity.
This is the singularity.
Epistemia is the singularity. Being inundated by plausible or even AI-checked information that we cannot verify is the singularity.
What can go wrong?
The most important human skill in the age of AI will be the ability to distinguish truth from falsehood.
We need to preserve this skill.
*
Paper where we introduce epistemic literacy in the first reply
33
Tauseef Khan retweeted
I don’t think I’m exaggerating when I say this: Inkling 2 Mini might be the inflection point for local ai.
1M context length, natively multimodal, only 20GB, AND as good as Qwen 3.8 27B
Some of you thought that Inkling 2 Small was too big at 594B parameters. We heard you: Inkling 2 Mini is a looped transformer with just 26B parameters, small enough to run on your laptop, powerful enough for agentic work with a score of 20.1 on AA Index.
thinkingmachines.ai/news/ink…
Tauseef Khan retweeted
Breaking News, and if true utterly astounding: Anthropic claims their AI has proved the Riemann Hypothesis, a towering target in mathematics for nearly two centuries, with profound ramifications.
Massive achievement!
🚨 JUST IN: Turkish-American mathematician Levent Alpöge, working with Claude, claims to have proved one of the hardest problems in mathematics: the Generalized Riemann Hypothesis!
The Riemann Hypothesis is one of the seven Millennium Prize Problems.
The announcement includes a massive Lean formalization, which, when written out in English, runs to over 250,000 pages! 🤯
If verified, this could be the most significant mathematical breakthrough achieved with AI to date.
59
Tauseef Khan retweeted
🚨 JUST IN: Turkish-American mathematician Levent Alpöge, working with Claude, claims to have proved one of the hardest problems in mathematics: the Generalized Riemann Hypothesis!
The Riemann Hypothesis is one of the seven Millennium Prize Problems.
The announcement includes a massive Lean formalization, which, when written out in English, runs to over 250,000 pages! 🤯
If verified, this could be the most significant mathematical breakthrough achieved with AI to date.


































