Daniel Litt
@littmath.bsky.social
Assistant professor (of mathematics) at the University of Toronto. Algebraic geometry, number theory, forever distracted and confused, etc. He/him.
We thank the many mathematicians who proposed problems. We’re also grateful to our editorial board, consisting of @thomasfbloom.bsky.social, @littmath.bsky.social, and Dan Romik, who helped us vet each of the proposed problems. Here are some of their thoughts on the project overall.
Maybe too obvious to be worth saying, but: frontier models are now obviously superhuman at some mathematical tasks, including ones that the profession has, historically, rewarded with prestige etc.
toddler, who has started sprinting in a random direction any time I blink: daddy, the high chair protects me, from running away!
[some-subscribed-rss] New Post: AI Skeptics: Erdos’s Conjecture Was Just the Beginning (with Daniel Litt), by Cathy O'Neil, mathbabe https://mathbabe.org/2026/07/06/ai-skeptics-erdoss-conjecture-was-just-the-beginning-with-daniel-litt/
Academia on the Line is now live. Our inaugural episode features Professors Melanie Matchett Wood and Daniel Litt (@littmath.bsky.social) discussing OpenAI's result on the Erdős Unit Distance Problem and what AI may mean for mathematical research. academiaontheline.com
Academia on the Line
academiaontheline.com
One of my projects has taken a turn into really classical algebraic geometry and it’s such a different feeling from my usual mathematics. Kind of a grab-bag of beautiful but sporadic special objects: Cayley octads, various constructions with theta characteristics, etc.
toddler: daddy, maybe when I’m big you’ll shrink down. maybe!
Just added a 15th problem to problemsilike.com! This one is a "non-abelian" analogue of the variational Hodge or Tate conjectures.
3yo daughter: can I kiss you? me: sure kiddo, how come? 3yo: *peck on cheek* because I love you! I want a beard. me: how come? 3yo: because I love beard!
Very challenging to communicate responsibly about this (AIxMath) stuff!
TBH I was pretty torn about contributing to this. In the end I decided that writing something restrained was better than writing nothing.
arXiv math.CO Combinatorics@mathco-bot.bsky.social · 3mo agoNoga Alon, Thomas F. Bloom, W. T. Gowers, Daniel Litt, Will Sawin, Arul Shankar, Jacob Tsimerman, Victor Wang, Melanie Matchett Wood: Remarks on the disproof of the unit distance conjecture https://arxiv.org/abs/2605.20695 https://arxiv.org/pdf/2605.20695 https://arxiv.org/html/2605.20695
Noga Alon, Thomas F. Bloom, W. T. Gowers, Daniel Litt, Will Sawin, Arul Shankar, Jacob Tsimerman, Victor Wang, Melanie Matchett Wood: Remarks on the disproof of the unit distance conjecture https://arxiv.org/abs/2605.20695 https://arxiv.org/pdf/2605.20695 https://arxiv.org/html/2605.20695
@littmath.bsky.social's commentary, noting that the low hanging fruit here may be because humans were just stuck on the wrong approach, trying to prove it true instead of false. In that sense I wonder how much the model benefits from all of those failed attempts not being part of the training corpus
mr. TIM@timkellogg.me · 3mo agoA general purpose (nothing special) internal OpenAI model solved one of the most famous previously unsolved problems in discrete geometry The solution involved far too many decisions for a human to feasibly explore openai.com/index/model-...
And we have our first proposed solution, to problem #6! I think it is quite likely to be correct, though I am still checking details. I would characterize the solution (produced by GPT 5.5 Pro, prompted by @thomasfbloom.bsky.social as "literature search plus epsilon"). See my preregistered comments:
![General remarks
Following the terminology of [A04b], this question asks for a computation of the "Frobenius amplitude" of an ample vector bundle in positive characteristic. The notion is motivated by the algebraic proof of Kodaira vanishing in [DI87], due to Raynaud.
The main result of [A04b] shows that, given 𝑋 in characteristic zero, ℰ be an ample vector bundle on 𝑋 and 𝒢 a coherent sheaf on 𝑋, there exists a spreading out of the data (𝑋,ℰ,𝒢) such that modulo almost all primes, the vanishing in the problem statement holds. In positive characteristic, the desired vanishing is only known under strong liftability assumptions [L19], namely that rk(ℰ) <char(𝑘), and that 𝑋 admits a lift
˜
𝑋
to 𝑊2(𝑘) such that ℰ(𝑝𝑁) lifts to
˜
𝑋
for some 𝑁 >0. While this yields some pleasant applications the hypotheses are likely not optimal.
Formalizability
I am not sure if this statement can be formalized given the current state of MathLib.](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:wvexgwndckqyszwviyuoeibi/bafkreicdljvmqfhm4cfffcb73y3guprhes3w5kmq6ojhgtt4b7wavsmdo4)
Daniel Litt@littmath.bsky.social · 3mo agoNew project: problemsilike.com, a website collecting open problems that I, personally, like, with comments on their context, difficulty, and interest.
New project: problemsilike.com, a website collecting open problems that I, personally, like, with comments on their context, difficulty, and interest.
I'm facilitating the FrontierMath: Open Problems workshop in Toronto. If you're a research mathematician in the area I encourage you to apply!
Epoch AI@epochai.bsky.social · 3mo agoJoin us for in-person workshops to develop problems for FrontierMath: Open Problems! We are seeking highly interesting unsolved problems from research mathematics whose solutions can be verified programmatically. These are hard to find. Come take a crack at it! Link below.
New paper just dropped, joint with Thomas Krämer and Marco Maculan. It's about a (somewhat mysterious, to me) connection between cubic threefolds and the exceptional Lie group E_6. 1/n
enjoy that sunset while you can—soon a swarm of superintelligent AI agents will be able to appreciate it more rapidly and efficiently than you ever could
The lines on a cubic threefold are parametrized by a (complex) surface, whose real points are pictured below in the case where the cubic threefold is cut out by the equation x_0^3+x_1^3+x_2^3+x_3^3+x_4^3=0. You can play with it yourself here: chocolitt.github.io/fermat_fano_... 6/n
Thomas Kr\"amer, Daniel Litt, Marco Maculan $E_6$-local systems from cubic threefolds https://arxiv.org/abs/2604.20970
New paper just dropped, joint with Thomas Krämer and Marco Maculan. It's about a (somewhat mysterious, to me) connection between cubic threefolds and the exceptional Lie group E_6. 1/n
Just submitted my 32nd paper to arXiv! Teaser: chocolitt.github.io/fermat_fano_...
Fermat Fano Surface
chocolitt.github.io
my wife: *brings toddler a bowl of berries* me: isn’t mommy so nice? toddler: i’m nice too! *gives me a berry* me: yes, you’re nice too toddler: *shakes head* i JUST SAID that
Toddler learned that I am lactose-intolerant and is concerned about it. About 3 times a day, she asks, “Daddy, does cow’s milk make you sick?” “Yes.” And then I get to hear a new list of things she really likes that might be dairy: “Does ice cream make you sick? Pizza? Kiwis?”
On today's episode of @scifri.bsky.social, @littmath.bsky.social and I had a conversation about vibe proving and AI for mathematics. Check it out here: www.sciencefriday.com/segments/cou...
Move over, vibe-coding. Vibe-proving is here for math
A few years ago, ChatGPT couldn’t do simple arithmetic. Now, some experts say that AI could make mathematicians obsolete.
sciencefriday.com
