> But we know that any person who uses AI is likely to improve at what they do. Do we?
Interesting post. The First Proof experiment really showed us the near future of AI/math interactions, some impressive success, but also lots of extremely hard to verify text, misformulated lean "proofs" etc. but…
Obviously not
NYTimes: https://www.nytimes.com/2026/02/07/science/mathematics-ai-pr...
This is a very interesting contribution to the AI/math space. I hope it can be seen by nonmathematicians interested in this. The mathematicians involved are quite well known (Martin Hairer is a Fields medalist). See…
Peer review should be disrupted, but doing peer review via social media is not the way to go.
Yes this is the standard proof of infinitely many primes but note that my prompt asked for infinitely many even primes. The point is that GPT would take the correct proof and insert "even" at sensible places to get…
This also can be observed with more advanced math proofs. ChatGPT 5.2 pro is the best public model at math at the moment, but if pushed out of its comfort zone will make simple (and hard to spot) errors like stating an…
My favorite early chatgpt math problem was "prove there exists infinitely many even primes" . Easy! Take a finite set of even primes, multiply them and add one to get a number with a new even prime factor. Of course,…
It's helped, but it's not correct that mathematicians are scoring major results by just feeding their problems to gpt 5.2 pro, so the OP claim that mathematicians are just playing off AI output as their own is silly.…
I think a more realistic answer is that professional mathematicians have tried to get LLMs to solve their problems and the LLMs have not been able to make any progress.
I think "pretty soon" is a serious overstatement. This does not take into account the difficulty in formalizing definitions and theorem statements. This cannot be done autonomously (or, it can, but there will be serious…
Somewhat unrelated but can anyone recommend a way to edit the text of a PDF using LLM? Something like AI + acrobat pro?
Very cool! This is one of the more impressive "AI does math" writeups I have seen.
Amazing how many different areas of math Jim Simons made serious contributions to, before turning to finance.
Assuming you have some math background but no Lean background: https://adam.math.hhu.de/#/g/leanprover-community/nng4
I think that (most) mathematicians were not that interested in formal proof until quite recently (as opposed to computer scientists), and most of the interest in lean has been self-reinforcing, namely there is a…
If it was a sure thing, why publish the paper they did? Why not just solve NS?
Citation based metrics are much more prevalent in physics than in math (at least in the US and most countries in Europe). When compared with physics, my impression is that mathematics has the tradition "slow, long term"…
The key is that mathematicians in the US and most parts of Europe do not count citations. So this is not really an issue.
This article does not seem to quite convey the experience of a pure mathematician. Yes, citation fraud is happening on an apalling scale, but no it is not a serious issue for mathematicians. The problem of AI generated…
It seems like we should switch from the academic model to a system of funding based purely on vibes.
This problem is solved by software like gradescope. Makes grading extremely fast and much more consistent (because of easy rubric adjustments on the fly). This is more concerning a STEM exam, admittedly I don't know how…
The commenters lamenting this trend presumably have not given a takehome assignment to college students in recent years. The problem is huge and in class tests are basically the only way to test if students are…
> But we know that any person who uses AI is likely to improve at what they do. Do we?
Interesting post. The First Proof experiment really showed us the near future of AI/math interactions, some impressive success, but also lots of extremely hard to verify text, misformulated lean "proofs" etc. but…
Obviously not
NYTimes: https://www.nytimes.com/2026/02/07/science/mathematics-ai-pr...
This is a very interesting contribution to the AI/math space. I hope it can be seen by nonmathematicians interested in this. The mathematicians involved are quite well known (Martin Hairer is a Fields medalist). See…
Peer review should be disrupted, but doing peer review via social media is not the way to go.
Yes this is the standard proof of infinitely many primes but note that my prompt asked for infinitely many even primes. The point is that GPT would take the correct proof and insert "even" at sensible places to get…
This also can be observed with more advanced math proofs. ChatGPT 5.2 pro is the best public model at math at the moment, but if pushed out of its comfort zone will make simple (and hard to spot) errors like stating an…
My favorite early chatgpt math problem was "prove there exists infinitely many even primes" . Easy! Take a finite set of even primes, multiply them and add one to get a number with a new even prime factor. Of course,…
It's helped, but it's not correct that mathematicians are scoring major results by just feeding their problems to gpt 5.2 pro, so the OP claim that mathematicians are just playing off AI output as their own is silly.…
I think a more realistic answer is that professional mathematicians have tried to get LLMs to solve their problems and the LLMs have not been able to make any progress.
I think "pretty soon" is a serious overstatement. This does not take into account the difficulty in formalizing definitions and theorem statements. This cannot be done autonomously (or, it can, but there will be serious…
Somewhat unrelated but can anyone recommend a way to edit the text of a PDF using LLM? Something like AI + acrobat pro?
Very cool! This is one of the more impressive "AI does math" writeups I have seen.
Amazing how many different areas of math Jim Simons made serious contributions to, before turning to finance.
Assuming you have some math background but no Lean background: https://adam.math.hhu.de/#/g/leanprover-community/nng4
I think that (most) mathematicians were not that interested in formal proof until quite recently (as opposed to computer scientists), and most of the interest in lean has been self-reinforcing, namely there is a…
If it was a sure thing, why publish the paper they did? Why not just solve NS?
Citation based metrics are much more prevalent in physics than in math (at least in the US and most countries in Europe). When compared with physics, my impression is that mathematics has the tradition "slow, long term"…
The key is that mathematicians in the US and most parts of Europe do not count citations. So this is not really an issue.
This article does not seem to quite convey the experience of a pure mathematician. Yes, citation fraud is happening on an apalling scale, but no it is not a serious issue for mathematicians. The problem of AI generated…
It seems like we should switch from the academic model to a system of funding based purely on vibes.
This problem is solved by software like gradescope. Makes grading extremely fast and much more consistent (because of easy rubric adjustments on the fly). This is more concerning a STEM exam, admittedly I don't know how…
The commenters lamenting this trend presumably have not given a takehome assignment to college students in recent years. The problem is huge and in class tests are basically the only way to test if students are…