Daniel Litt

24 Jan 2025

I want to explain in down-to-earth terms what this paper is about, since it ultimately boils down to what I think are some really concrete and fundamental questions. 1/n

arxiv math.AG@arxiv-math-ag.bsky.social

Yeuk Hay Joshua Lam, Daniel Litt Algebraicity and integrality of solutions to differential equations https://arxiv.org/abs/2501.13175

my toddler just asked me to draw her a picture, and then when I was done she hanged it on the fridge

honestly felt great

extremely concerning that OpenAI has made 3 sign errors—an odd number—upsetting this delicate balance

Daniel Litt@littmath.bsky.social

that the mathematical literature as a whole is so reliable comes down to the fact that, over time, we have made an even number of sign errors

quantamagazine.orgIs AI the End of Math As We Know It? | Quanta MagazineMathematicians are facing the sudden shift with grief, anger, and a desperate search for fresh ideas: “If we don’t adapt, there’s just no more math in 50 years.”

Good article here (in which I am quoted). As always with questions in headlines, the answer to the question in this headline is "no." www.quantamagazine.org/is-ai-the-en...

cmsa.fas.harvard.eduSummit on PhD Math Education in the Age of AI - CMSAOn September 17–18, 2026 a group of 24 mathematicians met at Harvard to grapple with the changing landscape for mathematics PhD programs in the age of AI. We produced recommendations given […]

that the mathematical literature as a whole is so reliable comes down to the fact that, over time, we have made an even number of sign errors

Daniel Litt reposted

emilyriehl.bsky.social

@emilyriehl.bsky.social

I was part of a group that met last week to try to propose recommended changes to the structure of math PhD programs in an age of AI. Our report, together with a collection of related resources, is now available here: cmsa.fas.harvard.edu/aimathphd_su...

Daniel Litt reposted

theHigherGeometer

@highergeometer.mathstodon.xyz.ap.brid.gy

Getting AI tools to help produce artifacts like this, from @littmath , is I think a nontrivial good application: https://www.daniellitt.com/fermat_fano_real_mesh_web.html To my mind this is analogous to how murmurations in number theory were discovered by […] [Original post on mathstodon.xyz]

Daniel Litt reposted

proofsandprompts.comA beginning for mathematicsDaniel Litt, professor at the University of Toronto Three years ago, AI systems could not reliably add two numbers. A year ago, internal models at OpenAI and DeepMind received the equivalent of a g…

I've written an essay on how I think the mathematics profession should adapt to highly capable AI systems. It's hosted here on "Proofs and Prompts": proofsandprompts.com/2026/09/14/a... though you should also feel free to complain/comment on my website here: daniellitt.com/blog/2026/9/...

Daniel Litt reposted

Joel Dodge

@weakthinker.bsky.social

18 Sept

Replying to @colin-fraser.net

Daniel Litt the wise AI commentator? I remember when it was Daniel Litt the S-tier shit poster. Just one more thing AI has taken from us.

Daniel Litt reposted

Colin

@colin-fraser.net

18 Sept

Replying to @colin-fraser.net

this probably feels so good if you're @littmath.bsky.social

Daniel Litt reposted

River_Tam

@rivertam.bsky.social

16 Sept

"Whatever our goals are, we’ve operationalized them primarily through proving theorems. Almost all papers or PhD theses have a main theorem, and ostensibly a proof of it. But it should be clear that the goal of mathematics is not to prove theorems..." proofsandprompts.com/2026/09/14/a...

Some actual prescriptions:

Anyway, please let me know what you think!

On pushing buttons and what we can and can't do:

On what we want to preserve:

The essay is mostly written for my colleagues but I hope non-mathematicians will get something out of it too. A couple excerpts...

Daniel Litt reposted

youtube.comDaniel Litt | Working with LLMs to do high quality mathYouTube video by Harvard CMSA

I gave a talk at CMSA on AIxMath on Wednesday. The title ended up a bit unrelated to the content--I rewrote the talk at the last minute for obvious reasons, and talked a bit more about my views on the future of the profession than I expected to. Link: www.youtube.com/watch?v=0wL8...

youtube.comDaniel Litt | Working with LLMs to do high quality mathYouTube video by Harvard CMSA

My goal is to spend less time talking about AI and more time talking about math for the next few months, but they pulled me out of retirement for one last job.

Incidentally I think the Q&A starting around minute 52 is arguably more interesting than the talk itself.

youtube.comDaniel Litt | Working with LLMs to do high quality mathYouTube video by Harvard CMSA

If you're not a mathematician I encourage you to skip the middle part (about math, about minutes 15-34) and just watch the beginning and end.

Daniel Litt reposted

theseouldan.bsky.social

@theseouldan.bsky.social

youtu.beDaniel Litt | Working with LLMs to do high quality mathYouTube video by Harvard CMSA

For a much more positive - and frankly much more informed - view than mine on LLMs in maths, this from @littmath.bsky.social is very good youtu.be/0wL8NlhxXcU?...

There's been some recent suggestions that recent AIxMath successes have relied on stealing ideas from mathematicians' work in progress. While we can't know for sure what happened, I think the evidence for this is very weak. 1/n

While I know many people at the labs care about science and the norms of good science, it seems to me that these norms were sacrificed as soon as there was a hint that there was a (frankly pretty inconsequential, though mathematically cool) prize to be won. 11/n

It makes me worry how they will behave if something more substantial is at stake, and what principles will be sacrificed for that. 12/n, n=12

That there is so much jockeying for credit between the labs, and people at the labs, suggests to me that the people closest to the technology might think it still matters. 9/n

On a related note, I find the race dynamics between the labs disturbing. 10/n

That some people are motivated by this does not make their work less important, and credit and prestige have historically incentivized good work. Of course I wish that we were all perfect truth-seekers and cared about this less. 7/n

Will credit continue to matter so much as model capabilities increase? I'm not sure; I very much hope not. I think this is more a question about the shape of society than the shape of model capabilities. 8/n

The weaker claim that PDFs or announcements put out by labs and lab employees often do not adequately attribute or credit other work seems correct to me. How important is it? 5/n

I wish that science was free from concerns of human ego but it's not. (Maybe it soon will be.) Top scientists mostly forgo substantial monetary compensation for their work, and are instead primarily compensated in credit. 6/n

Whatever one thinks of the labs, this is exceedingly dangerous to the health of profession; if human mathematics is going to survive we have to face facts about what the models can do. 3/n

And we should also celebrate exciting mathematics, even as we're concerned about mathematicians and mathematical expertise; who are these results for, if not us? 4/n

It seems to me that the primary function these claims have is to deny or downplay recent growth in model capabilities. 2/n

Daniel Litt reposted

22 July

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.

math.brown.edumath.brown.edu

30 Aug

Very pessimistic short story about AI and Math (among other things) by Richard Schwartz at Brown University, called "The Fate of the Riemann Hypothesis": www.math.brown.edu/reschwar/Sto...

30 Aug

(The worries the story expresses are not my worries, but it's entertaining nonetheless.)

Daniel Litt reposted