
Wake me when they prove subconvexity of the symmetric square 🥱

Math NSF Postdoc @MIT, Prev: Postdoc @AarhusUni, Math PhD @Princeton

Wake me when they prove subconvexity of the symmetric square 🥱

2027 and 2029 are the first twin prime years since 1997 and 1999 and the last until 2081 and 2083. If you have kids in '27 and '29 they can be twin prime siblings or 'twins' for short

I have a number of problems in mind whose proofs, if they go the way I expect them too, could really advance analytic number theory and be useful to the number theory community. But it seems the $200 home model version of these LLMs is ill-equipped for solving such problems (as am I perhaps lol) as @OpenAI has just demonstrated that the price tag is potentially hundreds to thousands of times as much as me or any other mathematician could ever reasonably pay. Is a frontier lab now the best place to be a mathematician?

Maybe this is cope but it still seems that AI progress in mathematics follows human progress. You need some sufficiently smart idea before models have a chance of finishing off the rest. On the other hand I worry this disincentivizes the sharing of ideas until it’s all published

Feel like these AI companies need like math publicists or something. It seems there’s been a lot of misses in announcing and working on results lately. Maybe in general it’d be good to develop some kind of ROE for how they’re approaching outstanding math problems

I don't repost mathematics if I do not personally feel sure of its correctness, so I did not yet comment on the purported construction of a complex structure on S^6. However, I wanted to add some contextualization because I don't agree with the social media narrative that is developing. 0. Of the AI breakthroughs so far it's the closest to my interests, although still pretty far. I am not grieving, I am interested to learn more about it! I don't know if it's correct but on cursory glance it seems quite plausible. 1. If correct, this would be a major development. I believe it is a central problem in its field (complex differential geometry), like the Unit Distance Conjecture was for combinatorial geometry. Perhaps even more famous names are attached to this field, leading some people to say it's even more prestigious, but I prefer not to think in such terms. 2. There are two writeups, a 2-page one and a 100+-page one. Both are terrible and seem deliberately constructed to be inhumane. Fortunately, one can fight AI with AI: I asked Claude/GPT/Gemini to collaborate on making a more readable and enlightening version, which you can find at overleaf.com/read/frddxmbnqb…. To make this, I prompted the models to write a document in the style of one of my own surveys. I haven't checked the mathematics, but it seems less forbidding than the original document at least. 3. This is a great opportunity for an expert to write an illuminating account of the construction, explain the key ideas and their proper historical context, and summarize the takeaways and lessons learned for the mathematics community. 4. If I see such an account, I would be happy to arrange for the author to give a lecture series in Berkeley. If this lecture series is also wonderful, I would go further and advocate for the author to have a position at Berkeley (if desired). For contrast, I see little value in having colleagues who are good at being the first to prompt AI for a solution. 5. There are some intriguing differences with the previous artifacts from Claude. The announcement seems coy about the provenance of the solution - there is a vague allusion to Claude, but which model(s) did it come from, and what was the human involvement? Why is it hosted on a personal website instead of the Anthropic one? We have seen some instances of AI results that are so good, the human authors seem to begrudge credit. I keep a list of such instances. The standard should be to disclose the role of AI clearly and transparently. Better to be honest than go through life under a cloud of suspicion. 6. Although this may solve one of the more famous problems of mathematics, there are still whole fields that are only just starting to feel acceleration. I'm not aware of any AI breakthroughs in my areas of research, for example. When those come, as I suspect they will before long, I look forward to studying the insights!
SITUATION EXPLAINED: Levent Alpöge just did it again, and mathematicians are grieving. @__alpoge__: "Please welcome to the world a beautiful new geometric object, to do with a problem I’ve always loved. claude really contains multitudes:D Does S^6 admit a complex structure? Yup" • The paper runs ~108 pages and constructs an explicit complex 3-fold diffeomorphic to S⁶. Some are calling it a Fields Medal level insight if a human had done it @doomslide: "I want to be happy about this but after reading and digesting the construction I think this is quite sad. It tells us very little about complex vs generalized complex structures. We had a prestigous, elegant problem that motivated people thinking about this and now it's gone." @boazbaraktcs's essay the same day: "Some mathematicians have been meeting this change with excitement, while others with grief. I share both sentiments." @theojaffee: "If you're a mathematician, Terence Tao is smarter than you. Does that mean it is not worthwhile for you to do mathematics because there exists a mathematician who's better than you? No, of course not."

We will still need people to digest and communicate the AI proofs to other humans. I think the point is more to learn things rather than to know things are true. If aliens came down and had a proof of RH it would be meaningless if we couldn’t explain it to others. AI is the same
I am now really worried about what math research will look like at the end of my PhD. OpenAI’s recently announced results are getting dangerously close to my research area, and I can already hear people telling me "did you ask an LLM" every time I ask a question

Fable, I’ve seen what you have done for others… Have been trying to get an answer out of this thing for weeks but it just keeps hitting its max length


Wow London culture is so different. Was in a Waterstones and overheard two friends looking for the green book and found a copy of Sutton & Barto in the tech section. US Barnes and noble could never


Something pretty strange just happened. Some random user’s question to ChatGPT appeared in my chat logs on my account. They’re asking about a color. At first I thought someone got into my account and asked a random question, but the answer seems to suggest the model has memory for this user being into art history which is a topic I’ve never brought up. Can other users chats randomly appear in one’s account?

Full clip is 1:43:00 - 1:46:03 in this youtube link youtube.com/watch?v=o9QqdX2O…

Really fascinating moment where @GlennLoury describes feeling an extreme sense of "stereotype threat". On the heels of earning a PhD in econ at MIT, he finds himself at NU with peers who would go on to win the nobel prize. He had great ambition and maybe could have kept up with those peers, but will never know because he was too afraid to stick with it. "The feeling of fear would've been there no matter what but the intolerable character of 'If I fail, I'll be the black guy who failed' was something that I just couldn't live with" Crazy that these are exactly the same thoughts I'm having after finishing math PhD at princeton, some 45 years later. Perhaps the regret of not knowing is much worse than the feeling of being the black guy who failed. youtube.com/clip/Ugkxcn7OTEF… via @YouTube

Y'all got me reading an AI math proof on a saturday night, on valentine's day no less 😭😅

I'm devastated. This is a massive loss for the world of chess. 😢
The Naroditsky family shares the sad news of Daniel’s unexpected passing. Daniel was a talented chess player, educator, and beloved member of the chess community. We ask for privacy as the family grieves.