Sign in

Huw Day

@huwwday.bsky.social
89 followers 125 following 80 posts

Research Associate in Digital Health at the VIVO Hub for Enhanced Independent Living (www.thevivohub.com) at the University of Bristol Co-organiser of the Data Ethics Club (dataethicsclub.com) Recovering mathematician he/him

PostsRepliesMedia
Reposted by Huw Day
Emily Hunt @emily.space · 20h
translation: "The world should accept some bad things happening for the benefits of my personal fortune"
47213
Reposted by Huw Day
Dr Abeba Birhane @abeba.blacksky.app · 02/10/2026
grant proposal: the prevalence of x on y is unknown and my proposal will explore this using method z reviewer: there is no empirical evidence that method z shows the prevalence of x on y
3421
Reposted by Huw Day
Andreas Kirsch @blackhc.bsky.social · 07/01/2025
Ever wondered why presenting more facts can sometimes *worsen* disagreements, even among rational people? 🤔 It turns out, Bayesian reasoning has some surprising answers - no cognitive biases needed! Let's explore this fascinating paradox quickly ☺️
824283
Reposted by Huw Day
Sam Power @spmontecarlo.bsky.social · 30/09/2026
nice to see it framed so clearly
1454
Reposted by Huw Day
Dr Abeba Birhane @abeba.blacksky.app · 29/09/2026
unsolicited brag: dug into the data -> ai -> surveillance pipeline in class this morning and received a standing ovation at the end of class. genuinely did not expect it. now i’m in tears. first time for everything
1634042
Reposted by Huw Day
Tom Gauld @tomgauld.bsky.social · 24/09/2026
My latest @newscientist.com cartoon. many more here: www.newscientist.com/author/tom-gauld/
Image: A graph with lots of red dots clustered together in a curve. The are all looking upwards at a single red dot wearing sunglasses. 

Caption: Mike was an outlier and would probably be excluded from the study, but all the other data points secretly thought he was really cool.
332878739
Reposted by Huw Day
Dr Abeba Birhane @abeba.blacksky.app · 22/09/2026
well, today in a class of ~150 CS students, only a few raised hands when asked if they trust AI 🤷🏾‍♀️🤷🏾‍♀️
18512
Reposted by Huw Day
William B. Fuckley @opinionhaver.bsky.social · 22/09/2026
what is 'Latent Space' and how do LLMs get answers from it? The Jargon can be confusing, but the exaplanation is actually relatively simple: It is Hell. The Machine is getting knowledge from demons.
3085995
Reposted by Huw Day
Shannon Vallor @shannonvallor.bsky.social · 17/09/2026
Sorry, the harmful, unsafe things done *since the last Summit*
172
Reposted by Huw Day
Shannon Vallor @shannonvallor.bsky.social · 17/09/2026
Truly astonishing that all any world leader can think to do is keep inviting the same guys into a room to give their considered advice on how we can protect ourselves from all the harmful, unsafe things deliberately done by the guys in the room
bbc.co.uk
King Charles warns of 'existential danger' of AI falling into wrong hands
The comments were made at a summit in Ayrshire with participants from AI giants including Nvidia, OpenAI and Anthropic.
610627
Reposted by Huw Day
Sam Power @spmontecarlo.bsky.social · 16/09/2026
(from www.its.caltech.edu/~matilde/AIm..., links to .pdf)
1136
Reposted by Huw Day
Terence Tao @teorth.bsky.social · 11/09/2026
A group of 25 Fields Medalists, including myself, have made a joint declaration on Math and AI: mathandai.org . We welcome additional signatories. See also this article in the Economist announcing the declaration: www.economist.com/science-and-...
mathandai.org
Declaration — Math and AI
Read the declaration and add your name.
422054930
Reposted by Huw Day
David Picard @davidpicard.eurosky.social · 08/09/2026
Remember folks, if it leaves your computer, it's no longer "your data"
39820
Reposted by Huw Day
Lauren Klein @laurenfklein.bsky.social · 25/08/2026
After far too many AI-generated final projects, and my own uncertainty about what students really need to know about NLP these days, I decided to overhaul my Text as Data class so that it's structured around a single project: training up a C19 LLM from scratch. Here we go! github.com/laurenfklein...
github.com
GitHub - laurenfklein/dsci340-fall26: Notebooks and other course materials for Emory DSCI 340 (Fall 2026 - Klein)
Notebooks and other course materials for Emory DSCI 340 (Fall 2026 - Klein) - laurenfklein/dsci340-fall26
28211
Reposted by Huw Day
Dr Abeba Birhane @abeba.blacksky.app · 22/08/2026
omg
2795
Reposted by Huw Day
Evangelos Kazakos @ekazakos.bsky.social · 30/07/2026
@iclr-conf.bsky.social There is a discrepancy/ ambiguity regarding the paper submission deadline. iclr.cc/Conferences/... and iclr.cc/Conferences/... say 25th of September while iclr.cc/Conferences/... says 16th of September. A clarification would be highly appreciated! 🙏
iclr.cc
2027 Dates and Deadlines
031
Reposted by Huw Day
Christian Wolf @chriswolfvision.bsky.social · 24/07/2026
This is fine ...
57710
Reposted by Huw Day
Alex Hern @hern.bsky.social · 21/07/2026
Don’t like this openai.com/index/huggin...
openai.com
OpenAI and Hugging Face partner to address security incident during model evaluation
OpenAI and Hugging Face share early findings from a security incident during AI model evaluation, highlighting advanced cyber capabilities and lessons for defenders.
115617
Reposted by Huw Day
Naomi Saphra @nsaphra.bsky.social · 20/07/2026
😭
Peter Henderson
@PeterHndrsn
Several people with no technical background told me they submitted to NeurIPS using agents for all coding—stating they didn't understand what the agent did, nor underlying code. 

Asked how they could trust the code/results, they said they assume reviewers will catch any errors.
49618
Reposted by Huw Day
Christian Wolf @chriswolfvision.bsky.social · 17/07/2026
In 2026 the accepted papers are the ones which were reviewed by the same LLM which the authors had used to check the paper before submission.
1131
Reposted by Huw Day
Gary Stevenson @garyseconomics.bsky.social · 14/07/2026
Watch the documentary: www.youtube.com/watch?v=V6hb...
youtube.com
Britain’s Inequality Problem...And How I’d Solve It | How To Get Filthy Rich | Channel 4 Docs
YouTube video by Channel 4 Documentaries
48428
Reposted by Huw Day
Benn Jordan @bennjordan.bsky.social · 10/07/2026
I found a hot stock tip for you.
4740423
Reposted by Huw Day
Jean Golding Institute @jgibristol.bsky.social · 09/07/2026
Job opportunity for Bristol Uni PhD students! We are recruiting a new team of PhD students for the Ask-JGI helpdesk to work from October 2026 until the end of September 2027! Find out more: jeangoldinginstitute.blogs.bristol.ac.uk/2026/07/06/a...
jeangoldinginstitute.blogs.bristol.ac.uk
Ask-JGI helpdesk is recruiting PhD students! – Jean Golding Institute News
133
Reposted by Huw Day
Andrew Sissons @acjsissons.bsky.social · 28/06/2026
My daughter said to me on Thursday “this is the hottest day of my life” and honestly the Homer/Bart “so far” meme has never been so bitter
356765
Reposted by Huw Day
Ketan Joshi @ketanjoshi.co · 25/06/2026
An analysis of 85 British news programmes during a May 2026 heatwave found only 4 mentioned fossil fuels as the cause explicitly. I suspect it'll be the same proportion for this current heatwave. What is the point, if the primary cause isn't stated clearly? climatenewstracker.org/during-the-h...
Linking the heatwave to climate change
Where the climate link was made, it was often made well.


Made with Flourish • Create a chart
On PM on 26 May, Carlo Buontempo of the Copernicus Climate Change Service explained why Europe is warming faster than any other continent. On Sky News Today on 25 May, correspondent Barnaby Papadopoulos told viewers that an extreme weather event like this one is now 30 times more likely than it would have been since pre industrial times.
47526
Reposted by Huw Day
Emily Hunt @emily.space · 25/06/2026
as an astronomer, I can safely tell you that stars are really fucking hot, and we should be doing everything in our power to protect our planet from the star next to us
713323
Reposted by Huw Day
TMLR Published Papers @tmlr-pub.bsky.social · 19/06/2026
DINOv3 Oriane Siméoni, Huy V. Vo, Maximilian Seitzer et al. Action editor: Jingcai Guo openreview.net/forum?id=2NlGyqNjns #visual #features #supervised
0154
Reposted by Huw Day
Coleen Murphy @ctmurphy1.bsky.social · 17/06/2026
one anecdote: I was searching for a link to one of my old papers on Google, and the automated Gemini summary attributed all of my lab's work to my husband
1041711907
Reposted by Huw Day
Prof Stephen Serjeant @stephenserjeant.bsky.social · 18/06/2026
Don’t trust LLMs with your literature searches. Horrible gender biases as found in this paper (from later in the thread below) arxiv.org/abs/2508.02740 🧪🔭
arxiv.org
Who Gets Cited? Gender- and Majority-Bias in LLM-Driven Reference Selection
Large language models (LLMs) are rapidly being adopted as research assistants, particularly for literature review and reference recommendation, yet little is known about whether they introduce demogra...
19148
Reposted by Huw Day
Rachel Coldicutt @rachelcoldicutt.bsky.social · 18/06/2026
I did a roundtable with some PhD students yesterday on the current existential problem of UK tech policy: it doesn't matter how much evidence you take to the table, it doesn't matter how credible the research is, you will have no impact if you don't promise AI will deliver the moon on a stick
47421
Reposted by Huw Day
Rachel Coldicutt @rachelcoldicutt.bsky.social · 18/06/2026
In the cold light of delay I feel quite distressed by the active in seriousness of this kind of grandstanding when taken side-by-side with the fact that the Home Office is rolling out technology that is not for for purpose www.wired.com/story/facial...
wired.com
The UK Will Scan Asylum-Seekers’ Faces for Age Checks—Despite Knowing the Tech Is Flawed
Internal Home Office tests of age-verification technology show the risks of life-altering errors. It’s moving forward anyway.
24711
Reposted by Huw Day
Jeff Sharlet @jeffsharlet.bsky.social · 16/06/2026
At the end of the term I asked my college creative wriing students to submit anonymous thoughts on AI. No real surprises: Mood ranges from resignation to despair, capitulation from embittered erosion of standards to total, feelings of betrayal from deep to furious. 1/
7455231988
Reposted by Huw Day
Mel Andrews @bayesianboy.bsky.social · 17/06/2026
When utilized in literature review, LLMs consistently 1. fail to mention female authors in female-led literatures, 2. insist that men are more influential or more heavily cited when this is contradicted by objective citation counts, and 3. attribute women’s work to hallucinated male scholars.
8153392302
Reposted by Huw Day
Dr Abeba Birhane @abeba.blacksky.app · 17/06/2026
dead funny but also painfully accurate
28620
Reposted by Huw Day
It's ME(Jaime) @exceedhergrasp1.bsky.social · 17/06/2026
A source on this: arxiv.org/abs/2508.02740
arxiv.org
Who Gets Cited? Gender- and Majority-Bias in LLM-Driven Reference Selection
Large language models (LLMs) are rapidly being adopted as research assistants, particularly for literature review and reference recommendation, yet little is known about whether they introduce demogra...
32613
Reposted by Huw Day
Emily Hunt @emily.space · 05/06/2026
I am SO sick of headlines like this. The public sees "AI" as being LLMs/ChatGPT, not an ML algorithm - and articles like this mislead the public into thinking that LLMs are discovering novel things. It's a cool thing! I'm excited! But it is lazy reporting to suddenly use "AI" as such a broad term
Headline that reads 'World-first' vaccine designed by artificial intelligence
30507161
Reposted by Huw Day
depths of wikipedia @depthsofwikipedia.bsky.social · 29/05/2026
89155574912
Reposted by Huw Day
Dr Abeba Birhane @abeba.blacksky.app · 08/07/2025
a couple of hours before my keynote, I went through an intense negotiation with the organisers (for over a hour) where we went through my slides and had to remove anything that mentions 'Palestine' 'Israel' and replace 'genocide' with 'war crimes' 1/
381446683
Huw Day @huwwday.bsky.social · 31/05/2026
Chatting stats with @spmontecarlo.bsky.social is associated with airport queues moving faster (can't prove causality tho)
130
Reposted by Huw Day
Sam Power @spmontecarlo.bsky.social · 29/05/2026
graphic design remains my passion sites.google.com/view/newcast...
0233
Reposted by Huw Day
The Onion @theonion.com · 20/05/2026
OpenAI Announces Construction Of New Data Center On Top Of Sick Child
12197742141
Reposted by Huw Day
Karen Hao @karenhao.bsky.social · 20/05/2026
On the one-year anniversary of EMPIRE OF AI, I am so, so excited to announce The AI Resist List, a new project that documents examples of resistance to the AI empires around the world 😍 airesistlist.org
The gorgeous AI Resist List homepage
4525271122
Reposted by Huw Day
Dr Abeba Birhane @abeba.blacksky.app · 23/03/2026
New paper from @aial.ie! @harshp.com, Dick Blankvoort, Adel Shaaban, @sashamtl.bsky.social & me We analysed 6 GenAI ToS--finding missing info, major power imbalances & user obligations that are impossible to meet without violating the terms arxiv.org/abs/2603.18964 & aial.ie/research/ter... 1/
arxiv.org
Terms of (Ab)Use: An Analysis of GenAI Services
Generative AI services like ChatGPT and Gemini are some of the fastest-growing consumer services. Individuals using such services must accept their terms of use before access, and conform to these ter...
2193112
Reposted by Huw Day
Sam Power @spmontecarlo.bsky.social · 18/05/2026
Waheyy
36010
Reposted by Huw Day
Emily Hunt @emily.space · 18/05/2026
there is nothing more satisfying than starting the day with an O(n²) problem and finishing the day with it optimized into an O(n) problem
2883
Reposted by Huw Day
Dr Abeba Birhane @abeba.blacksky.app · 18/05/2026
Not just lobbying! our new paper maps 27 mechanisms of regulatory capture used by Big AI + 11 narrative framings that rationalise capture. “Big AI’s Regulatory Capture: Mapping Industry Interference and Government Complicity” will be presented at #FAccT2026 next month arxiv.org/abs/2605.068... 1/
Title: Big AI’s Regulatory Capture: Mapping Industry Interference and Government Complicity

abstract: Over the past decade, the AI industry has come to exert an unprecedented economic, political and societal power and influence. The well-functioning of regulatory and oversight structures and processes that govern the industry thus have paramount ramifications for everything from fostering public trust in systems marketed as AI, the credibility of scientific knowledge, educational and healthcare services and products, information ecosystems, the environment, rule of law and integrity of democratic process. In this paper, we first develop a taxonomy of mechanisms enabling capture to provide a comprehensive understanding of the problem. Grounded in design science research (DSR) methodologies and extensive scoping review of existing literature and media reports, our taxonomy of capture consists of 27 mechanisms across five categories. We then develop an annotation template incorporating our taxonomy, and manually annotate and analyse 100 news articles. The purpose behind this analysis is twofold: validate our taxonomy and provide a novel quantification of capture mechanisms and dominant narratives. Our analysis identifies 249 instances of capture mechanisms, often co-occurring with narratives that rationalise such capture. We find that the most recurring categories of mechanisms are Discourse & Epistemic Influence, concerning narrative framing, and Elusion of law, related to violations and contentious interpretations of antitrust, privacy, copyright and labour laws. We further find that Regulation stifles innovation, Red tape and National Interest are the most frequently invoked narratives used to rationalise capture. We emphasize the extent and breadth of regulatory capture by coalescing forces — Big AI and governments — as something policy makers and the public ought to treat as an emergency.
7377221
Reposted by Huw Day
Peoria Bummer @peoriabummer.bsky.social · 17/05/2026
It would be VERY interesting to study what biases might be present when completing a list of authors.
091
Reposted by Huw Day
Dr. Casey Fiesler @cfiesler.bsky.social · 17/05/2026
I've spent a LOT of time fixing .bib in LaTeX by hand. The most common issue is that BibTeX pulled from something like Google Scholar has missing information. The most common missing thing is author first names... and requests for fixes from an LLM *consistently* makes them up.
313018
Reposted by Huw Day
Dr. Casey Fiesler @cfiesler.bsky.social · 17/05/2026
Particularly in the context of (appropriate) crackdowns on LLM-generated papers (e.g., on arXiv) and more and more tales of fabricated citations, two reminders: (1) LLMs do not "cite their sources" (2) Using an LLM to generate, complete, or fix LaTeX code can result in fabricated citations
6419125
Reposted by Huw Day
Jia-Bin Huang @jbhuang0604.bsky.social · 17/05/2026
**Modern Transformer - Complete Guide** Interested in learning the recent advances in transformers? After 14 videos, I've finally completed this series! 🥳🥳🥳 Check out the course here: www.youtube.com/playlist?lis...
0325