Sign in

Alexander Kasprzyk

@amkasprzyk.bsky.social
478 followers 364 following 1.1K posts

kasprzyk.work Algebraic Geometry • Combinatorics • Mirror Symmetry • Machine Learning Mathematician at the University of Warwick. Editor-in-Chief for “Experimental Mathematics”. Member of EPSRC’s Math Strategic Advisory Team.

PostsRepliesMedia
Alexander Kasprzyk @amkasprzyk.bsky.social · 2h
I hadn’t realised the US brain-drain was this bad. Crickey.
010
Alexander Kasprzyk @amkasprzyk.bsky.social · 2h
I can’t fathom the appeal of reading a work of fiction or poem generated by an LLM. I think most people would agree. Similarly LLM-generated mathematics leaves me cold. Sure, you could lie about the source, and I would probably be fooled. That’s not the point. I value the human presence. #MathSky
1100
Reposted by Alexander Kasprzyk
The New York Review of Books @nybooks.com · 11h
“What we are witnessing is a spectacle of computational power staged to acquire market share, and with it control of a landscape that extends well beyond mathematics.” —Dan Rockmore
nybooks.com
The Work of Math in the Age of Artificial Reproduction | Dan Rockmore
There are seven questions in pure mathematics deemed important enough by the Clay Mathematics Institute, a nonprofit foundation devoted to “furthering the
0173
Reposted by Alexander Kasprzyk
Social Media Lab @socialmedialab.ca · 29/09/2026
"China has expanded overseas travel restrictions for top AI professionals in private firms to include the families of key personnel, further tightening measures designed to prevent the outflow of critical know-how and information to the U.S." www.japantimes.co.jp/business/202...
japantimes.co.jp
Beijing broadens travel curbs to encompass family of top AI talent in China
That further restricts the freedom of China’s most influential tech talent — a category that encompasses entrepreneurs, researchers and executives.
178
Alexander Kasprzyk @amkasprzyk.bsky.social · 5h
"AI's solution of Conjecture 9.1" by Jinyoung Park and Michel Talagrand. The proofs presented in this document were generated entirely by the #AI model GPT-6 Astra, and none of the authors claim any credit for them. This write-up is ... not intended for journal publication. arxiv.org/abs/2609.33644
arxiv.org
AI's solution of Conjecture 9.1
In an earlier paper, M. Talagrand made a number of daring conjectures. An AI model provided a very beautiful proof of one of them which we explain here.
100
Alexander Kasprzyk @amkasprzyk.bsky.social · 7h
…or, perhaps, a Fields Medal? Probably not the model itself, but quite possibly one of the mathematicians working in a large AI company such as OpenAI. #MathSky
110
Alexander Kasprzyk @amkasprzyk.bsky.social · 8h
www.theguardian.com/uk-news/2026...
theguardian.com
MI5 issues spy alert to UK universities over Chinese front company stealing research
Security service says academics should stop taking grants from and working with firm backed by China’s intelligence service
000
Reposted by Alexander Kasprzyk
London Mathematical Society @londmathsoc.bsky.social · 13h
The 2026-27 Round of LMS Early Career Fellowships are live, open to early career mathematicians in the transition between PhD and a postdoctoral position. Deadline is TOMORROW (1 October 2026). For further details ➡️ www.lms.ac.uk/grants/lms-e...
012
Reposted by Alexander Kasprzyk
Neil Chadborn @neilchadborn.bsky.social · 13h
Today & tomorrow are first days of strike as Uni council makes final decisions on massive cuts of courses &staff. Please show your #solidatity by sharing this, to encourage management to take our alternative financial proposals seriously. Compulsory redundancies not necessary!! @uonucu.bsky.social
02219
Alexander Kasprzyk @amkasprzyk.bsky.social · 13h
Perhaps the time has come to establish an alternative to the #arXiv, intended for depositing largely AI-derived results? An #aiXiv. Papers posted to the arXiv could still make use of #LLMs. The aiXiv would be intended for articles where the *majority* of the result originates from an #AI. #MathSky
A silly mockup of a logo for the aiXiv, based on the arXiv logo.
440
Alexander Kasprzyk @amkasprzyk.bsky.social · 29/09/2026
I participated in this survey back in December 2024. Looking at the results today, I see the average year respondents anticipated for #AI to "prove publishable math theorems" was around 2039. I don't recall what I guessed, which is probably for the best. #MathSky aiimpacts.org/wp-content/u...
aiimpacts.org
Advanced AI according to 1,580 researchers: uncertain, unsafe, and sooner than we thought
In December 2024, we ran a survey about the future of AI, reaching 1,580 researchers internationally who had published at any of six leading artificial intelligence (AI) venues, corresponding to a 10% response rate...
141
Alexander Kasprzyk @amkasprzyk.bsky.social · 29/09/2026
"Giulio Regeni, a 28-year-old Italian citizen, was in the second year of his doctorate researching Egyptian labour rights... His mutilated corpse was found beside a road several days later." Absolutely terrible.
130
Alexander Kasprzyk @amkasprzyk.bsky.social · 29/09/2026
I'm co-I on a grant whose PI is at a very wealthy UK university. For over 14 months that university has failed to transfer my share of the award. Yesterday I asked them "would it be helpful if I contacted the funder tomorrow and explained the situation?" Today the money is being transferred. #UKHE
1200
Alexander Kasprzyk @amkasprzyk.bsky.social · 29/09/2026
Many users of Muse are not only giving it *their* private data, they are giving it *your* private data. Muse is accessing emails and private message histories such as Apple's Messages. The confidentiality of every personal message or email you ever sent to a friend now depends on their #AI agent.
040
Reposted by Alexander Kasprzyk
Emmanuelle Chaze @emmanuellechaze.com · 29/09/2026
Right now, living in Kyiv feels like being in a giant cat & mouse game that the rest of the world watches with deep concern. Each day is a life lottery for everyone - you wake up from a horrific night to dive into a horrific day, and you try to go through it alive. And repeat. Every day. 1/
14014
Alexander Kasprzyk @amkasprzyk.bsky.social · 29/09/2026
For many disciplines, the cost of cutting-edge research can be prohibitively expensive. Leading researchers concentrate in dedicated high-profile institutes such as the Francis Crick Institute (a magnet for funding), or buy time on shared community resources such as the Large Hadron Collider.
The Francis Crick Institute in London, next to the British Library and opposite St Pancras train station. It's an amazing building.
250
Reposted by Alexander Kasprzyk
Lucy Gaynor @lucyjgaynor.bsky.social · 28/09/2026
Today, staff @ University of #Edinburgh are striking to protest recently announced cuts - incl. the forced closure of >90 courses. Principal Prof Sir Peter Mathieson accused the union of "weaponising" students. I find this enraging, so: a short (quite personal) thread. www.bbc.com/news/article...
bbc.com
Edinburgh University lecturers begin five-week strike over cuts
Members of the University and Colleges Union are walking out in part of a long-running dispute over job and course cuts.
14121
Alexander Kasprzyk @amkasprzyk.bsky.social · 29/09/2026
The ability to end these disruptive strikes lies in the hands of @uniofnottingham.bsky.social managers. The staff on strike are trying to save the university and the education of current and future students. #UCU #UKHE www.bbc.co.uk/news/article...
bbc.co.uk
Student: University of Nottingham strikes 'not what we signed up for'
The University of Nottingham said it was "disappointed" by the timing of industrial action.
0199
Reposted by Alexander Kasprzyk
Clément Canonne @ccanonne.github.io · 28/09/2026
From our friends at the School of #Mathematics and Statistics at #USyd: positions as 'Magma Research Associates / Magma Research Fellows to develop and maintain the Magma computer #algebra system." 2 positions, each for 3 years: apply by October 20 usyd.wd105.myworkdayjobs.com/en-US/USYD_E...
usyd.wd105.myworkdayjobs.com
Magma Research Associate / Magma Research Fellow
2 x full time, 3-year fixed term positions, located on the Camperdown Campus at the School of Mathematics and Statistics Opportunity to drive research in computational pure mathematics through the dev...
189
Alexander Kasprzyk @amkasprzyk.bsky.social · 28/09/2026
Already mathematicians have to decide whether they want to use their grant application to fund a postdoc, or instead to purchase #LLM tokens. Perhaps soon your department will offer you a choice between a PhD studentship or LLM tokens. These decisions will shape the future of our community. #MathSky
A picture of an abandoned lecture hall, intended to convey the choice that faces researcher between tokens and research students.
181
Alexander Kasprzyk @amkasprzyk.bsky.social · 28/09/2026
"Sylvester simplices: Triangulations and Ehrhart-theoretic aspects" by Jhon B. Caicedo, Federico Castillo, Martina Juhnke and Germain Poullot. #LatticePolytope #Combinatorics #MathSky arxiv.org/abs/2609.23710
arxiv.org
Sylvester simplices: Triangulations and Ehrhart-theoretic aspects
The Sylvester simplex $\mathsf{Sylv}_d^k$ is a $d$-dimensional lattice simplex with exactly $k$ interior lattice points. Sylvester simplices are conjectured to be the volume maximizers among all $d$-d...
010
Reposted by Alexander Kasprzyk
Jeffrey Shallit 🇺🇦 @shallit.bsky.social · 18/09/2026
A new risk for mathematicians: a colleague reports that immediately after they posted an abstract of their upcoming talk online, a student elsewhere used AI to derive proofs of the stated results and then posted it on the arxiv. My colleague hadn't posted to arxiv yet. Scummy. Watch out.
16219
Alexander Kasprzyk @amkasprzyk.bsky.social · 28/09/2026
The Ukrainian National Academy of Sciences in Kyiv has been attacked – in the middle of the day – by Russian drones. This is deliberate terrorism.
150
Alexander Kasprzyk @amkasprzyk.bsky.social · 28/09/2026
I worry that universities view the cost of providing access to an #LLM rather like an institutional subscription to Microsoft Office. They are unwittingly entering an arms race that only those organisations with very deep pockets can survive. #UKHE #MathSky
An image of a ancient Greek pavilion, intended to suggest the collapse of the current research-intensive university system.
160
Alexander Kasprzyk @amkasprzyk.bsky.social · 28/09/2026
“The surest defence against Evil is extreme individualism, originality of thinking, whimsicality, even — if you will — eccentricity. That is, something that can’t be feigned, faked, imitated.” —- Exiled Soviet poet and Nobel prizewinner Joseph Brodsky.
010
Alexander Kasprzyk @amkasprzyk.bsky.social · 27/09/2026
An engineer, a physicist, and a mathematician are at a high-school math event, and have been asked to explain a trigonometry problem to the students. #MathSky
XKCD cartoon #435, expressing the purity of mathematics compared with other subjects. An axis along the bottom represents "purity", with socialists followed by psychologists followed by biologists followed by chemists followed by physicists. There is then a vast gap, after which appears the mathematician.
1113
Alexander Kasprzyk @amkasprzyk.bsky.social · 27/09/2026
More tokens and better #LLM models in the hands of researchers translates into significant results. This is now pretty clear. There is no limit on the number of tokens we could consume. Science risks becoming a game for those with disposable wealth; the deepest pockets “win”. How can we avoid this?
4170
Alexander Kasprzyk @amkasprzyk.bsky.social · 26/09/2026
“Scoring Grant Applications with Large Language Models” “”” Whilst the correlations seem too weak to replace expert review at the final panel stage, LLM scores might help with the initial reviewing state... “”” arxiv.org/abs/2609.25327
arxiv.org
Scoring Grant Applications with Large Language Models
Purpose: Assessing grant applications is time-consuming and difficult, adding to the overall burden of academic peer review. Whilst funders are exploring whether AI can help, there is no published res...
100
Reposted by Alexander Kasprzyk
François Loeser @francoisloeser.eurosky.social · 26/09/2026
Jean-Pierre Serre's lecture on his 100th birthday is now available: www.youtube.com/watch?v=tNto...
youtube.com
Jean-Pierre Serre - Plaisir des mathématiques
YouTube video by Institut Henri Poincaré
03012
Alexander Kasprzyk @amkasprzyk.bsky.social · 25/09/2026
"Equivariant Log Concavity and the FI^#-Module Structure on H^i(Conf(n,\RR^d))" by Benjamin Homan. #ExperimentalMath #MathSky #Combinatorics #RepresentationTheory www.tandfonline.com/doi/full/10....
tandfonline.com
Equivariant Log Concavity and the FI♯-Module Structure on Hi (Conf(n,ℝd))
Gedeon, Proudfoot, and Young first conjectured that the graded Sn -representations H∙(Conf(n,Rd);Q) are strongly equivariantly log concave for even d. Later work of Matherne, Miyata, Proudfoot, and...
120
Alexander Kasprzyk @amkasprzyk.bsky.social · 24/09/2026
"A Szemerédi-Trotter Theorem in Arbitrary Fields" by Mark Lewko. """ Let k be a field of characteristic p>0. We prove that m points and n lines in k^2 determine O((mn)^{2/3}+m+n+mn/p) incidences. In characteristic zero the last term is omitted... """ #MathSky #Combinatorics arxiv.org/abs/2609.27023
arxiv.org
A Szemerédi-Trotter Theorem in Arbitrary Fields
Let $k$ be a field of characteristic $p>0$. We prove that $m$ points and $n$ lines in $k^2$ determine $O((mn)^{2/3}+m+n+mn/p)$ incidences. In characteristic zero the last term is omitted. The proof us...
010
Alexander Kasprzyk @amkasprzyk.bsky.social · 24/09/2026
#AI “triaging” of UKRI grants. Yikes. I completely appreciate the struggle they have finding timely reviewers, but surely there’s a better way than this? #UKHE
442
Alexander Kasprzyk @amkasprzyk.bsky.social · 24/09/2026
"Three Conjectures on the Łojasiewicz Exponent of an Isolated Hypersurface Singularity" by Szymon Brzostowski and Tomasz Rodak. #ExperimentalMath #MathSky #AlgebraicGeometry www.tandfonline.com/doi/full/10....
tandfonline.com
Three Conjectures on the Łojasiewicz Exponent of an Isolated Hypersurface Singularity
For an isolated hypersurface singularity f:(\CC^n,0_\rightarrow(\CC,0), we study the Łojasiewicz exponent ł(f) of f. We formulate three conjectures concerning: (i) attainment of ł(f) on coordinate polar curves, (ii) topological invariance of ł(f), and (iii) a local Newton-diagram formula for ł(f) in the Kushnirenko non-degenerate case. We compute ł(f) for all singularities of modality at most 2 in Arnold’s list and for several further families, obtaining systematic evidence in favor of these conjectures. We also find to what extent the collection of numerical invariants {modality⁡(f),ł(f),\mu(f),corank⁡(f),stabord⁡(f)} determines the stable equivalence class of f for the aforementioned singularities.
010
Reposted by Alexander Kasprzyk
London Mathematical Society @londmathsoc.bsky.social · 23/09/2026
Just 1️⃣ week left to apply for the Editor-in-Chief or Deputy Editor-in-Chief vacancy for our membership publication, the LMS Newsletter If you are an LMS member, and have sufficient time and commitment to the fulfil the role, please consider applying ➡️ www.lms.ac.uk/newsletterad...
011
Reposted by Alexander Kasprzyk
London Mathematical Society @londmathsoc.bsky.social · 22/09/2026
💐 Applications for the next round of Emmy Noether Fellowships are open! Grants of up to £10,000 are open to mathematicians with current or recent caring responsibilities. Deadline: Mon 30 November Further details and to apply ➡️https://www.lms.ac.uk/grants/lms-emmy-noether-fellowships
023
Alexander Kasprzyk @amkasprzyk.bsky.social · 22/09/2026
Public perception about the future of mathematics is an enormous challenge facing our community. We are currently not even remotely in control of the narrative. For example, the photo shows what’s on display in every newsagents and supermarket in the UK. Will future students be scared away? #MathSky
The latest issue of New Scientist magazine, with the front cover “How AI blew up maths”.
5288
Alexander Kasprzyk @amkasprzyk.bsky.social · 21/09/2026
I've been reading a lot recently about #CambridgeAnalytica. One thing that stands out is how gleefully prepared academics were to participate in what was clearly immoral and unethical work. E.g., a US medical school volunteered their database of millions of patients' genetic profiles. #AcademicSky
The logo of Cambridge Analytica: a network of nodes and lines arranged in the profile of a human brain.
1132
Alexander Kasprzyk @amkasprzyk.bsky.social · 21/09/2026
We all make mistakes. And the difference in style between LLM-generated text and human-authored text is slim, and will only get smaller. Having a distinctive voice helps, but it only gets you so far. Let's not fall into the unpleasant trap of witch hunts and policing. There are enough zealots.
000
Reposted by Alexander Kasprzyk
Iris van Rooij 💭 @irisvanrooij.bsky.social · 28/02/2026
📚 And some more advice from Mary E. Boole (1909, p. 33) which oddly reads as highly relevant for AI today … 👀 21/🧵
Take care not to become a parasite; do not lazily appropriate the results of other people’s labour, but learn and labour truly to get your own living. Take care that everything you possess, whether physical, mental, or spiritual, shall be the result of your own toil as well as other people’s; and remember that you are bound to pay, in some shape or way, everyone who helps you.
18119
Alexander Kasprzyk @amkasprzyk.bsky.social · 21/09/2026
"Mellin Transforms, Transfinite Diameter and Rational Approximations of Integrals" by Francis Brown. #ExperimentalMath #MathSky #NumberTheory www.tandfonline.com/doi/full/10....
tandfonline.com
Mellin Transforms, Transfinite Diameter and Rational Approximations of Integrals
We establish a higher-dimensional irrationality criterion for periods which are presented as Mellin integrals depending on many parameters. The criterion is stated as an upper bound on the multi-variate transfinite diameter of the image of the domain of integration under the Mellin arguments. Most of the paper is devoted to studying notions of transfinite diameter relative to very general multivariate Vandermonde matrices. As a proof of principle, we illustrate how this approach works with detailed computations in the case of a 5-parameter family of integrals for \zeta(2) on \mathcal{M}_{0,5}, the moduli space of curves of genus 0 with 5 marked points. This yields a “higher-dimensional” proof of the irrationality of \zeta(2), based on an upper bound for a certain kind of transfinite diameter associated to \mathcal{M}_{0,5}.
020
Alexander Kasprzyk @amkasprzyk.bsky.social · 21/09/2026
67 days of strike action at @uniofnottingham.bsky.social #UKHE #Nottingham
011
Alexander Kasprzyk @amkasprzyk.bsky.social · 19/09/2026
“[I]ncessant smartphone use has dramatically reduced attention spans, replacing ‘deep reading’ with scrolling, and pupil motivation has collapsed.” This is a genuine problem for everyone.
100
Reposted by Alexander Kasprzyk
Alexander Kasprzyk @amkasprzyk.bsky.social · 18/09/2026
The Warwick University Maths Houses are open to the public this Sunday as part of England's Heritage Open Days. If you're in the area, they're fun to see. @uni-of-warwick.bsky.social #MathSky #Heritage www.heritageopendays.org.uk/submission-e...
heritageopendays.org.uk
University of Warwick Maths Houses
Built in 1969, the Grade II* listed status Maths Houses won the 1970 RIBA Architecture Award. Designed by Bill Howell, with distinctive curved walls and integrated blackboards. Look around, draw on…
2106
Alexander Kasprzyk @amkasprzyk.bsky.social · 18/09/2026
The Warwick University Maths Houses are open to the public this Sunday as part of England's Heritage Open Days. If you're in the area, they're fun to see. @uni-of-warwick.bsky.social #MathSky #Heritage www.heritageopendays.org.uk/submission-e...
heritageopendays.org.uk
University of Warwick Maths Houses
Built in 1969, the Grade II* listed status Maths Houses won the 1970 RIBA Architecture Award. Designed by Bill Howell, with distinctive curved walls and integrated blackboards. Look around, draw on…
2106
Alexander Kasprzyk @amkasprzyk.bsky.social · 18/09/2026
The picture on the poster is fascinating. There appears to be just one woman, who has six fingers. And a guy with two forearms coming out of a single elbow. Not to mention the man talking to a mirror image of himself. And everybody is the same height, with black hair. Odd.
3155
Alexander Kasprzyk @amkasprzyk.bsky.social · 18/09/2026
Final one for now: an interview with Karim Adiprasito in "Enumerative Combinatorics and Applications". I surely met Karim when he was doing his PhD at FU Berlin in the early 2010s, but sadly I don't have any recollection of this. #Combinatorics #MathSky ecajournal.haifa.ac.il/Volume2022/E...
ecajournal.haifa.ac.il
Interview with Karim Adiprasito
Karim Adiprasito completed his undergraduate studies at TU Dortmund in 2010. He obtained a Ph.D. from Free University Berlin in 2013, under the supervision of Gunter M. Ziegler. He has been an Assistant Professor at the Einstein Institute for Mathematics Hebrew University of Jerusalem since 2015, promoted to full professor in 2018. Professor Adiprasito has given numerous invited talks in conferences and seminars. He received several prizes, for example, European Prize in Combinatorics, in 2015, for his work in discrete geometry; New Horizons Prize for Early-Career Achievement in Mathematics, associated with the Breakthrough Prize in Mathematics, in 2019; Prize of the European Mathematical Society in 2020, and is the 2021 Hadamard Lecturer.
020
Alexander Kasprzyk @amkasprzyk.bsky.social · 17/09/2026
He gives no indication that the authors were warned their responses would be used in a public article; he gives no indication that this went through any ethics committee checks; and he fed the unpublished papers to an LLM, which is a big no-no, especially given recent concerns. #LLM
191
Alexander Kasprzyk @amkasprzyk.bsky.social · 17/09/2026
I'm delighted that my article "Kawamata bounds for Fano threefolds and the Graded Ring Database", joint with Gavin Brown and Miles Reid, has appeared in the Nagoya Mathematical Journal. #MathSky #AlgebraicGeometry www.cambridge.org/core/journal...
cambridge.org
Kawamata bounds for Fano threefolds and the Graded Ring Database | Nagoya Mathematical Journal | Cambridge Core
Kawamata bounds for Fano threefolds and the Graded Ring Database - Volume 261
0120
Alexander Kasprzyk @amkasprzyk.bsky.social · 17/09/2026
That this was achievable is surprising to me. But what also surprises me is the extent of the damage done by Iran that somehow didn’t make it into the news at the time. Presumably because we grossly underfund our traditional news media, and largely ignore reporting on non-Western (social?) media.
020