Applications are open for visiting PhD @compscibristol.bsky.social @bristoluni.bsky.social in 2026 - DL 29 Jan Would you like to work with any of the Faculty working in Machine Learning and Computer Vision #mavi as part of our summer of research at Bristol program? uob-mavi.github.io/Summer@MaVi....

Why we need to tax the super rich more. Their wealth is growing MUCH faster than everyone else's. And they're using it to outbid you for real resources: housing, assets, healthcare, education and everything else of real value.

Ok, I have a strong opinion on this, perhaps you can convince me otherwise: if you need an LLM to understand a paper for reviewing, perhaps you should not be a reviewer. And your argument is saved time, then perhaps you planned to not read all the details. Then again, you should not be a reviewer.

The US social media vetting for visas will be devastating for scientific and journalistic conferences, fellowships etc. No global organisation can seriously consider holding an international conference in the US while this policy exists.

the talk deliverer, back in action (from this morning's talk at Manchester CS)

New paper hot off the press www.nature.com/articles/s41... We analysed over 40,000 computer vision papers from CVPR (the longest standing CV conf) & associated patents tracing pathways from research to application. We found that 90% of papers & 86% of downstream patents power surveillance 1/

Artificial intelligence research has a slop problem, academics say: ‘It’s a mess’

I LOVE THIS SO MUCH. Please please please please stop giving me news articles about blog posts about AI that include images of blue data and robot hands. betterimagesofai.org

The invited NeurIPS talk by Zeynep Tufekci. Very interesting'

NeurIPS 2025 Best papers are out: - Artificial Hivemind: The Open-Ended Homogeneity of LLMs ... - Gated Attention for LLMs ... - 1000 Layer Networks for Self-Supervised RL ... - Why Diffusion Models Don’t Memorize ... (+ Runner-ups) blog.neurips.cc/2025/11/26/a...

Quite fun

Physician who was no longer at the hospital but was on the email list had Otter ai installed and his bot “attended” the meeting, generated a transcript, and sent it to all 65 people on the email list, 12 of whom also no longer worked at the hospital.

You can still sign up to the course for free, worldwide (it runs again in the Spring and Summer) www.futurelearn.com/courses/sust... #sustainabledevelopment #SDGs #sustainabledevelopmentgoals #COP30 #ESD #sustainability @bristoluni.bsky.social @cabot-institute.bsky.social @biltonline.bsky.social

🧵 With over 20 runs of the course & 21,000 learners, the Sustainable Futures course we made a few years ago is a runaway success. Why does it matter? It looks at local and international case studies covering the environmental, social, economic and cultural pillars of sustainable development.

A new paper with Bogdan Georgiev, Javier Gomez-Serrano, and Adam Zsolt Wagner: "Mathematical exploration and discovery at scale" arxiv.org/abs/2511.02864. Further discussion is at terrytao.wordpress.com/2025/11/05/m...

Data Feminism -- @kanarinka.bsky.social @laurenfklein.bsky.social Data Feminism is the best introduction to issues of justice and data science. I assign chapters of this book in every course I teach. This book lays bare how some data-driven systems empower people and others oppress.

I appreciate that this is technically a comprehensible sentence (and pretty benign in terms of how complex the concepts are), but the density of jargon did hit me with that vague feeling of "am I having a stroke".

Roughly speaking, the prompt to the discussion was "big things are happening elsewhere, how are we going to respond to this in our field?", and this speaker gave a very calm answer to the tune of "it is okay to still focus on existing challenges, it is not essential to respond to this other stuff".

efficiency. The amount of work that goes into applying to these grants is insane -- the academics writing it, the academic colleagues commenting, the Professional Service teams finetuning and sorting out all the costings etc. Surely that time could be used more efficiently *for actual research*?

Enjoying reading kellerjordan.github.io/posts/muon/ to learn more about the Muon optimizer!

Really nice primer on double-descent and the bias-variance trade-off I'm impressed by the depth that Welch Labs consistently manages to pack into their videos without sacrificing the storytelling for a popular science audience www.youtube.com/watch?v=z64a...

Some degree of self-awareness here (I think!):

An interesting knock-on effect of various recent hype cycles at the intersection of LLMs and Maths is that I've found some motivation to care a bit more about what people really mean by "open problem".

very convenient that { function, gradient, Hessian } come in such neat alphabetical order

Ever since I made a video about Fourier Transforms, one of the most requested topics on the channel has been its close cousin, the Laplace Transform. I've been having a lot of fun animating a mini-series about this topic, and the main part is now out. youtu.be/j0wJBEZdwLs

What PhD do I need to get and from which institution to fully understand git?

I'm looking for advice on LLM proof assessments on content related to neural networks. If I had more time with the students (I teach an intro to the topic across 3 hours, but they are masters level students), I would have them read a paper and replicate it in PyTorch.

Some fun news: We're hiring! Lectureship (analogous to Assistant Prof.) in Statistical Science and AI, in the School of Mathematics, University of Bristol. Closing date 13 October 2025; link below! www.bristol.ac.uk/jobs/find/de...

Sam Altman isn't keen on you reading my book. EMPIRE OF AI is based on 300+ interviews, 7 yrs covering AI, and my time as the first reporter who got extensive access to OpenAI. I sought OpenAI's perspective throughout. For months they said it was coming. It never did. Pre-order here: empireofai.com.

Much of Euclid’s Elements is easily misunderstood. Some proofs seem to have logical gaps. Some constructions seem pointless, others seem needlessly convoluted. Each of these provides a window into how the ancient Greeks thought about math and the philosophical role that geometry played.

How to name your method: a comprehensive flow chart

The experiences of reading well-written papers about middling results and poorly-written papers about fascinating results are wildly different, in this (and in other) regard(s).

Perhaps a bit orthogonally to the main point here, I have certainly had the experience of reading something out of { duty / necessity / etc. }, and found it remarkable how this context can amplify all sorts of frustrations about the style, even those which could otherwise be quite minor.

When emailing for PhD positions, it's a smart move to use LLMs to avoid typos and grammatical mistakes ... but check for hallucinations! "I am particularly drawn to your recent work on multi-view video understanding and your development of datasets .. (e.g., the Something-Something dataset). "

@hypestudies.bsky.social conference starts tomorrow, choo choo! I'm very excited to share our preprint with @cianooo.bsky.social and Joshua Moon: "Can buzzwords create infrastructures? Reflecting on the Water-Energy-Food Nexus approach" . zenodo.org/records/1699...

Quite funny:

UK government report based on internal civil service trial finds Copilot doesn’t increase productivity, and indeed makes Excel take longer and with more errors, and requires Powerpoint users to have ‘corrective action’ applied to their outputs. www.theregister.com/2025/09/04/m...

New video about a piece by the modern artist Sol LeWitt, and the group theory behind it. youtu.be/_BrFKp-U8GI

Not based at Pittsburgh but want to hear more about stress management during a #PhD? I've got you covered! Earlier this year I gave a similar webinar for @biochemsoc.bsky.social which is available here: youtu.be/2T6Mgoz04XI?... P.S. I've not watched it, please don't tell me if it's embarrassing! 🙈

Reading @savvyscientist.bsky.social 's blog in the lead up to my PhD viva did a lot for calming my nerves, always recommend listening to what he has to say!

In reading research papers, it's somewhat inevitable to occasionally have the wrong mental pronunciation of an author's name. Easily corrected, not a big deal, etc. What I find funny is that for me, the most common source of making such mistakes is to erroneously assume that the author is French.

Guest video 3/5 while I'm on leave is now up! It's by a former SoME winner, covering key ideas in statistical mechanics to create a simple and discrete model mirroring the behavior of a fluid transitioning between a liquid and gaseous state. Enjoy! youtu.be/itRV2jEtV8Q

dear funding agency please give me money to prove what we already know but with more decimal places

The jobs that are most vulnerable to LLMs aren't the jobs that LLMs can do well; they're the jobs that managers neither respect nor understand and will readily accept permission to replace. *waves* www.bloodinthemachine.com/p/ai-killed-...

Effective captioning:

Very sweet annotations:

This was such a rewarding gig!

I want to advertise some relatively recent work which I really like, and have been fortunate to play a small role in. The paper is titled "A New Proof of Sub-Gaussian Norm Concentration Inequality" (arxiv.org/abs/2503.14347), led by Zishun Liu and Yongxin Chen at Georgia Tech.

The government will tax the middle class out of existence instead of trying to tax the vast wealth of the very rich.