Daniel Litt

Assistant professor (of mathematics) at the University of Toronto, visiting Harvard Fall 2026. Forever confused. Algebraic geometry, number theory, etc. He/him.

[Daniel Litt @littmath]

New paper with Josh Lam, about which I'm really excited! I want to try to briefly explain what the point is in this thread.

[Daniel Litt @littmath]

@benediamond i think the general expectation among mathematicians was that the hodge conjecture was hard...

[Daniel Litt @littmath]

@aleshu_ber IMO it's pretty misleading to clip the video here. While I think Altman's description of the day zero/day one reaction etc. are incorrect, he goes on to describe a very different reaction from the one you leave it at.

[Harvard Department of Mathematics @HarvardMath]

Thanks to Daniel Litt from
@UofT for being a part of this year's #BottLectureSeries! He spoke about "The arithmetic of differential equations" and "Non-abelian motives" to an eager audience of mathematicians at Harvard's Science Center/Zimmer Hall.
#harvardmath

[Daniel Litt @littmath]

@mathgeek234 @nihilunbounded @slopdropz I think it’s fine to cry in public if you like, but also that not all mathematicians are crying—some are excited about the possibilities for new mathematics.

[Daniel Litt @littmath]

@benediamond @plain_simon Just narrowly commenting on “any that are interactive in the sense that the camera can be moved, panned and zoomed”

[Daniel Litt @littmath]

@benediamond @plain_simon Thanks! FWIW e.g. Desmos has allowed such visualizations for many years, e.g. https://t.co/8kpLqiVjOt

[Daniel Litt @littmath]

@benediamond @plain_simon https://t.co/xDeg43bktJ

[Daniel Litt @littmath]

@alz_zyd_ I believe this is possible, and in fact that pure math is such a field :P.

[Daniel Litt @littmath]

@ludwigABAP I think it means something like "give me a billion dollars."

[Daniel Litt @littmath]

@PI010101 I think it’s good advice to experiment with AI. But there’s some tension between maximizing this and learning to do mathematics oneself at a high level (since doing the former forces you to spend much of your time reading bot output rather than thinking on your own).

[Daniel Litt @littmath]

@max_spero_ This person is a crank.

[Daniel Litt @littmath]

@GWHayduke97 @zetalyrae I wonder what the procedure by which “we determine to our own satisfaction that it’s fully aligned” is in this scenario.

[Daniel Litt @littmath]

@alz_zyd_ wat

[Daniel Litt @littmath]

@docmilanfar Seems to me one could do worse than channeling Gauss!

[Daniel Litt @littmath]

Gauss had a personal seal(!) emblazoned with the motto “Pauca sed Matura” ("Few, but Ripe"). Worth remembering as quantity becomes cheaper.

[Daniel Litt @littmath]

@outis_000 How so? It’s quite possible to get a good job in math with just a few excellent publications.

[Paul Novosad @paulnovosad]

If someone asks you to write a reco letter and you plan to do it with AI, you should disclose it to them first.

It can come across as very lazy and low-effort and will harm the candidate.

If you want your people to succeed, don't do this.

[Daniel Litt @littmath]

some of you seem not to believe we want humans in the loop except as recipients of vertical video

[Daniel Litt @littmath]

@ObhishekSaha @BuenoEnfurecido @dan2nic OK, but we might develop norms that make it easier to apportioning credit accurately or according to our goals. This is not really an argument against such norms.