Proofchecker

[Proofchecker @proofchecker]

@janarhertz @scottnarmstrong Or: desk reject slop.

[Proofchecker @proofchecker]

@mattwridley Because your passport has an EXPIRY date.
You were already pretty close to the answer.

[Proofchecker @proofchecker]

@tiozzo_giulio Of course AI usage is no reason to reject a paper. Unreadability of a paper or larger parts of a paper is a very good reason for rejection. Whether the paper is unreadable because of AI usage isn't important.

[Proofchecker @proofchecker]

@bramiozo @ValerioCapraro And "announcing" a result without paper is accepted behaviour. It's what Thurston did and he was celebrated for it.

[Proofchecker @proofchecker]

@bramiozo @ValerioCapraro Mathematicians do not have to be on Twitter. They are here of their own choice. What Elliot did was share with us what was on his desk (and being completely honest about it). Nothing wrong with that. Way better than "announcing" a result with the paper coming years later if ever.

[Proofchecker @proofchecker]

@alijlari @AlexKontorovich "and if you actually understand it" -> "and if the reader with the proper background knowledge could actually understand it with a reasonable amount of effort".

[Proofchecker @proofchecker]

@AvitabileD @AlexKontorovich Reproducibility is not a thing in pure math. What matters is that the proof is correct and that the paper is sufficiently well written. How it came into existence shouldn't be important.

[Proofchecker @proofchecker]

@cuaternioX @AlexKontorovich All "what matters is the journey" argument are arguments against public funding of math.

Understanding is important, yes. But not the understanding in the mind of a single person. Understanding is clearly written text that others can read in the future to understand the result.

[Proofchecker @proofchecker]

@MichaelZaal8 @AlexKontorovich How about the ability to multiple 10-digit numbers in your head?

[Proofchecker @proofchecker]

@DynamicsSIAM @AlexKontorovich Bias etc. plays no role in mathematics.
Reproducibility plays no role in mathematics.
All that matters is that the proof is correct (and the paper is sufficiently well written and understandable).
Ideally every new paper comes with a Lean formalization to ensure correctness.

[Proofchecker @proofchecker]

@PeterVermont @AlexKontorovich So do you acknowledge your pocket calculator?

[Proofchecker @proofchecker]

@lflightning0815 @AlexKontorovich So if a result can be reproduced in 15 minutes by a prompt, it is not OK to publish the paper if the result was obtained using an LLM but it is OK to publish the paper if the author did not use an LLM?
That makes absolutely zero sense unless you see math as a jobs program.

[Proofchecker @proofchecker]

@torgeirlysen @AlexKontorovich Then your department totally misunderstands the function of a methodology section, which is to provide the information that readers/reviewers need to evaluate whether the results of the paper are valid and to reproduce them. There is no need for this in a pure math paper.

[Proofchecker @proofchecker]

@GOFIntelligence @AlexKontorovich in that case there is no need to "fix" the horrendous AI math writing and we should start looking for the best intermediate format that allows LLMs to exchange their mathematical thinking and conclusions.

[Proofchecker @proofchecker]

@GabrielDunin @NoahChrein John Grothendieck

[Proofchecker @proofchecker]

@ElliotGlazer All you did was share what you were doing. People who don't like that should just leave Twitter.

[Proofchecker @proofchecker]

@nucleotaid @gefiltefishy @animegfz Next time just do not start listening. Start collecting stamps.

[Proofchecker @proofchecker]

@alphabatcher @ValerioCapraro And why not.

[Proofchecker @proofchecker]

@AndersonEv37024 @highbrow_nobrow Sure but there is no coming back now.