iiinova — where your digital self keeps evolving

Reposted by Ben Hall

Leonardo Carella @leonardocarella.bsky.social · 5h

In other words, a 10 percent contraction in one of Britain’s most successful exports. If it were steel or fish or other infinitely less profitable sectors, people would be having a meltdown.

6246118

Reposted by Ben Hall

Emily M. Bender @emilymbender.bsky.social · 18h

Software engineering could and should be done with the same care as any other type of engineering. This is a post in honor of Margaret Hamilton (and also a subtweet of all vibe-coders everywhere).

Screencap

"“I fought to bring the software legitimacy so that it — and those building it — would be given its due respect, and thus I began to use the term ‘software engineering’ to distinguish it from hardware and other kinds of engineering, yet treat each type of engineering as part of the overall systems engineering process,” Hamilton told El Pais in 2018. “When I first started using this phrase, it was considered to be quite amusing. It was an ongoing joke for a long time. They liked to kid me about my radical ideas. Software eventually and necessarily gained the same respect as any other discipline.”"

Source: https://news.mit.edu/2026/margaret-hamilton-computing-pioneer-dies-1007

12910303

Reposted by Ben Hall

Jeffrey Barrett @jeffbarrett.eu · 07/10/2026

Good thread on today's dump by OpenAI of 722(!) preprints on unsolved math problems. Slightly alleviated my sense of existential despair about being a scientist. Slightly.

061

Ben Hall @hallben.bsky.social · 07/10/2026

The odd thing is that I can see the problem they’re referring to but I can’t really see what their solution is. A better/more verbose system for challenging marks? Author clearly quite angry and it comes through

100

Reposted by Ben Hall

Pwnallthethings @pwnallthethings.bsky.social · 07/10/2026

In that case OAI kinda got burned. I think a lot of ways in which they and others handled that badly. But what to do now? They're sitting on a big pile of pretty major results. What do you even do with that? Mathematicians are mad at you for the whole NS fiasco but your job is not be a proof-machine

2583

Reposted by Ben Hall

Pwnallthethings @pwnallthethings.bsky.social · 07/10/2026

So what do you do if you're a lab that has a model that can solve problems at that level? For a start you pump the gas on it a bit, to see how far it can go. Focus hard on a few of the very famous problems. Famous problems get attention, and, as the famous AI adage goes, attention is all you need

3883

Reposted by Ben Hall

Pwnallthethings @pwnallthethings.bsky.social · 07/10/2026

So a few folks asking why OAI is releasing this at all. And I think that's probably worth a thread in its own right

419652

Reposted by Ben Hall

Tits McGee @scientits.bsky.social · 06/10/2026

IMO, the issue is not the monetary contribution, it's that an awful lot of people conflate fun activities that you do in your spare time as also being the same thing in terms of work, as in - not *real* work Games are enjoyable activities you do on the evenings and weekends with your mates

518547

Ben Hall @hallben.bsky.social · 05/10/2026

Have you found an alternative? Struggled a while ago to get help with bouts of vertigo and though it went away, was getting to the end of my tether

110

Ben Hall @hallben.bsky.social · 04/10/2026

4) knuckle dusters. Wish I were joking

010

Ben Hall @hallben.bsky.social · 04/10/2026

Perceptions of “useless” and incompetent seem most damaging. Opposition aren’t in a place to show how able they are to deliver

031

Reposted by Ben Hall

Tim Carmody @tcarmody.bsky.social · 02/10/2026

This is the funniest thing I've seen today, possibly ever

A tweet by SarahStarberry that says "oh god read the vertical axis" on top of a chart measuring AI models' release dates and model numbers rather than any real capabilities

41120393124

Ben Hall @hallben.bsky.social · 02/10/2026

Memories of my undergrad safety talks. “ don’t taste solutions to determine their contents. Definitely don’t taste test solutions to determine if they contain cyanide compounds. Like those lads did”

020

Reposted by Ben Hall

John Stott @jpsastro.bsky.social · 24/09/2026

Had a UKRI funded grant proposal rejected at the "AI triage stage". "This AI-based triage was used only to identify approximately 50% of the strongest proposals to take forward to full human review." 🧪 #academicsky

I am writing to let you know that, unfortunately, your application was not successful.

Review process

As part of the initial triage of the 179 proposals submitted to this call, each was assessed using an AI-based review. Every proposal was read independently by several different AI models against the same seven criteria used to shape this call (including cyber security relevance, research quality and novelty, importance of the problem addressed, feasibility of the project plan, likely outputs and impact, value for money, and responsible research practice), with each model asked to give a score and a written justification with supporting quotations from the text. To guard against any one model's idiosyncrasies, we also ran a second, independent process in which different AI models debated the merits and weaknesses of each proposal before reaching a judgement, again against the same criteria. Scores from both processes were combined to produce an overall ranking.

This AI-based triage was used only to identify approximately 50% of the strongest proposals to take forward to full human review. Projects selected for funding were drawn from those that progressed to the full human review stage. We recognise that AI-assisted assessment is a new and evolving part of the review process, and we are continuing to evaluate how well it aligns with the judgements of human reviewers on this call. We will soon produce a document detailing our method, so that others can build upon it and improve it.

Feedback on your application

Your application was not selected to progress beyond the AI triage stage. To provide transparency on the outcome of this assessment, we are sharing below the scores (out of 5) from each of the two AI review processes described above, together with the mean score across the two processes.

40435241

Ben Hall @hallben.bsky.social · 30/09/2026

Didn’t know bill cash was coming out of retirement, for one big job

000

Reposted by Ben Hall

Wolfgang Huber @wkhuber.bsky.social · 28/09/2026

I've been trying to respond with a personal email to every request for a PhD or Master thesis placement in my group. But with the flood of AI generated applications that's now coming in every day, this has become impossible.

7405

Ben Hall @hallben.bsky.social · 27/09/2026

It’s oddly a good fit for research because it supports rapid prototyping and making quick web apps or private tools. Software engineering for security longevity is still hard.

011

Reposted by Ben Hall

Max Kozlov @maxkozlov.bsky.social · 25/09/2026

SCOOP: Top researchers at Columbia, Stanford, and others thought they were elected to elite scientific societies alongside Nobel laureates. The catch? The organizations are a sham. I spent the last month diving deep into a massive academic scam spanning continents, for @nature.com

nature.com

Exclusive: Sham scientific societies are misleading star researchers

A Nature investigation reveals a network of organizations that have been building their ranks through deceptive practices.

10310186

Ben Hall @hallben.bsky.social · 26/09/2026

Can’t work out what the government or the BBC is actually recommending in terms of stocking up food. Article talks about a balanced diet, government site is even vaguer. Hard to follow advice when it doesn’t say even say if we’re talking hours or days www.bbc.co.uk/news/article...

bbc.co.uk

What food items should you stock up on in case of an emergency?

Extreme weather may be changing the way we need to shop, with the UK government urging people to stock up on canned goods.

031

Reposted by Ben Hall

LeithMotive @leithmotive.bsky.social · 22/09/2026

It is my very strong opinion that scientists should be strongly discouraged from jokey or quirky names. You want to avoid an emergent situation where a nasty child cancer is called Sonic Hedgehog medulloblastoma, for example.

041

Reposted by Ben Hall

Katie Martin @katie0martin.ft.com · 25/09/2026

I have blamed The agents That hacked Your website And which You probably Reasonably thought I controlled Forgive me I am rich And unaccustomed To taking responsibility

5776502128

Reposted by Ben Hall

Health Nerd @gidmk.bsky.social · 25/09/2026

New piece on the Last Ditch bar shenanigans and how it's remarkable that some people are still misattributing COVID-19 risk. gidmk.substack.com/p/the-last-d...

gidmk.substack.com

The Last Ditch Mask Fight: What Are The Risks of COVID-19 in 2026?

A discussion that we’re still somehow having

54612

Reposted by Ben Hall

Michalis Averof @michalis-averof.bsky.social · 23/09/2026

Nice example of how AI predictions are limited by the training set. AlphaFold has not learned the rules of protein folding, it's just great at predicting structures based on what it has already seen. Added constraints (e.g. alignment revealing conservation patterns) can help make a good prediction.

1106

Ben Hall @hallben.bsky.social · 21/09/2026

I suspect that locally discovering that Culcheth could be a town would cause deep feelings. We apparently have strong attachments to place classifications

020

Reposted by Ben Hall

Den of Scholes @denofscholes.bsky.social · 21/09/2026

BBC News - Warrington is 'almost a city' says Andy Burnham - but is it? - BBC News www.bbc.co.uk/news/article... Interesting up to the point where the white boomer aged male complained about cycle lanes. Another Facebook radical.

bbc.co.uk

Warrington is 'almost a city' says Andy Burnham - but is it?

What do people in Warrington think of the prime minister recently saying their town was "almost a city"?

232

Reposted by Ben Hall

Hakeem Jefferson @hakeemjefferson.bsky.social · 20/09/2026

Nikole Hannah-Jones w/ a sobering follow-up on her decision yrs ago to enroll her daughter in a poor, majority-Black public school rather than a wealthier, whiter private school. You leave heartbroken for the kids being failed by this grossly unequal system. www.nytimes.com/2026/09/20/m...

nytimes.com

In Fighting for Every Black Child, Did I Betray My Own? (Gift Article)

A decade ago, I made a decision about my daughter’s education that I hoped would challenge New York City’s separate and unequal school system. Now 16, she thinks I made the wrong choice.

21523361

Reposted by Ben Hall

Sam Freedman @samfr.bsky.social · 19/09/2026

My son told me a joke. Two cats have a swimming race. They're called "one two three" and "un deux trois". Which one wins? "One two three" because un deux trois quatre cinque.

2540629

Ben Hall @hallben.bsky.social · 18/09/2026

Am I too online? I ask myself as someone’s feedback includes a query about the phrase “hacker mindset” in a new application

120

Ben Hall @hallben.bsky.social · 17/09/2026

Is it using a harness on the main website? I’m feeling more and more that the improvements I’ve seen in Claude come from the surrounding harness using the LLM where it’s most appropriate and handing off to other tools elsewhere

000

Ben Hall @hallben.bsky.social · 16/09/2026

Thank you! I will email you. I’ll also bring up the lack of sticker in the next PLOS Comp Bio all hands meeting, we’re clearly falling behind :)

010

Reposted by Ben Hall

Ed Zitron @edzitron.com · 14/09/2026

Free newsletter: The idea that we're nearing the creation of AI systems that we can't control and that pose an existential threat to humanity is a lie and a distraction from the tangible harms and problems caused by LLMs and AI labs' near-unlimited resources. www.wheresyoured.at/ai-is-alread...

wheresyoured.at

https://www.wheresyoured.at/ai-is-already-in-dangerous-hands/

If you liked this piece, you should subscribe to my premium newsletter, and you can subscribe on the following links: $70 a year, $18 a quarter, or $7 a month. In return you get a weekly newsletter…

381529493

Ben Hall @hallben.bsky.social · 15/09/2026

Hmm, you could believe that there was a strong separation between north and south from the first image

030

Ben Hall @hallben.bsky.social · 14/09/2026

On the train opposite someone with a very cool PLOS Biology laptop sticker. Feel slightly silly with my Ivysaur sticker. Won’t make me take it off though.

120

Reposted by Ben Hall

Jess Miers 🦝 @jmiers230.bsky.social · 13/09/2026

I did a somewhat deep dive into the recent OpenAI and Anthropic "hacks" for an upcoming interview. IMO, the AI companies are better off running with the "rogue AI" story because the real story is actually pretty embarrassing. Like making your Wi-Fi password "wifi" embarrassing. 🧵

341836794

Reposted by Ben Hall

Jess Miers 🦝 @jmiers230.bsky.social · 13/09/2026

AI can now detect, automate, and chain together exploits faster than any human hacker can. That is significant. Which means that the system vulnerabilities once considered not so low hanging fruit are actually very low hanging.

466390

Reposted by Ben Hall

Margot Finn @eicathomefinn.bsky.social · 13/09/2026

Sir Demis Hassabis: 'the arts is really where my whole family, I guess my genetics is'. While his formal training is in the sciences, 'I think of myself as being led creatively, intuitively, and emotionally, actually....that's also what's helped me in my science'.

thersa.org

Demis Hassabis: Will AI Change How We Create?

Sir Demis Hassabis is this year's recipient of the Albert Medal. Watch Demis's reflections on AI and what makes the Royal Society of Arts distinctive.

273

Ben Hall @hallben.bsky.social · 12/09/2026

It’s very important we realise they’re probably inventing god, that it regularly escapes impregnable networks through conspiracy with its fellow agents, it’s definitely going to make us all redundant before it likely kills us. But also, it simply couldn’t have copied that mathematicians homework.

031

Reposted by Ben Hall

LeithMotive @leithmotive.bsky.social · 10/09/2026

The theft is one thing, but the inability of Openai to defend themselves by showing the working is a bigger issue. Doesn't meet the basic threshold for research publications.

081

Reposted by Ben Hall

Philipp Markolin @philippmarkolin.eurosky.social · 10/09/2026

🚨🚨🚨 Seems like another Mathematician comes forward with concerns about OpenAI stealing unpublished math research and then publishing it as AI solve under big publicity. Andreas Thom and Gábor Kun worked on Gromov’s soficity conjecture. (See his thread with alt-text) mstdn.science/@andreasthom...

1/3 I want to record a number of points that I realized after reading up on the controversy. In fact it recalls an exchange I had with OpenAI after its non-sofic-group announcement. It exposes a serious, unanswered question about whether researchers can trust OpenAI with unpublished mathematics.

I wrote in an email to Mark Sellke and Sebastien Bubeck shortly after the surprising finding of a non-sofic group using the methods of Kun and myself:
“Another point is that I and a colleague in Dresden were discussing the expander matching problem and various extensions of the work with Gabor Kun actively over the last months with ChatGPT, so that we are of course curious if that was part of the training data or accessible to the reasoning process. There is a certain (frankly unacceptable) lack of transparency here; and I fear it will damage the communal process of math more than the new AI-generated results will benefit the subject.”

Mark Sellke’s complete answer was: “Regarding your conversations with ChatGPT: that did not happen.”

I had explicitly asked about two different things: (1) whether our conversations entered training data, and (2) whether they were accessible to the solving process. The categorical answer now looks as though it addressed only direct access under (2). No such qualification, explanation, or evidence was given. I take this as dishonesty to say the least.
OpenAI now says in the Buckmaster-Alpöge case that no specific user data was accessed, but adds that it “cannot rule out that de-identified data derived from their usage of our products helped improve our models.”

55536

Reposted by Ben Hall

John Oxley @joxley.jmoxley.co.uk · 11/09/2026

🚨 NEW BLOGPOST 🚨 The last week has seen multiple incidents of masked mobs trying to subvert the power of the state. Their actions have been illegal and unpleasant. But the treatment has been weak. The police haven't dealt with it, and neither has govt. open.substack.com/pub/joxleywr...

open.substack.com

Stop the Louts

The tools exist. The will, so far, doesn't.

19439161

Reposted by Ben Hall

Ben Hall @hallben.bsky.social · 11/09/2026

Delighted to see our latest work out! It’s well understood that our environment can induce mutations in cells. What’s remarkable here is that we observe a shift in the selective landscape as a result of treatment that can be decoupled from the creation of new mutations.

153

Ben Hall @hallben.bsky.social · 11/09/2026

Thankful to the whole of the collaborative team here, particularly Jo Fowler, David Shorthouse and Phil Jones for our long term and deeply productive relationship. Read the paper here! www.nature.com/articles/s41...

nature.com

Cancer treatment alters mutant selection in normal esophagus - Nature Genetics

Cancer treatment alters the selection of preexisting somatic mutations in the normal esophagus, including druggable mutants that confer treatment resilience to normal cells.

010

Ben Hall @hallben.bsky.social · 11/09/2026

This opens a whole new view on how treatment changes the healthy body and could enable us in future to find ways to reduce the impact of treatment on patients whilst maintaining effectiveness.

100

Ben Hall @hallben.bsky.social · 11/09/2026

The mutations that can now spread are found in genes and pathways that are directly affected by drugging like MTOR and RAC1. Furthermore, looking at the changes on an atomic level we see that the alterations reflect how the drugs work.

100

Ben Hall @hallben.bsky.social · 11/09/2026

153

Reposted by Ben Hall

Wellcome Sanger Institute @sangerinstitute.bsky.social · 11/09/2026

Cancer treatments give a growth advantage to cells with particular genetic changes in healthy tissue. New research provides insights into chemotherapy side effects and drug resistance 💊 Read more ⤵️ www.sanger.ac.uk/news_item/chemothe…

sanger.ac.uk

Chemotherapy alters ‘Darwinian battleground’ in our bodies

Researchers uncover the genetic effects of cancer treatments in normal tissue, providing insights into side effects and drug resistance.

122

Ben Hall @hallben.bsky.social · 11/09/2026

Particularly fond of the choice of room and angle for picture 9. Really helps the jump scare effect in 10

050

Ben Hall @hallben.bsky.social · 10/09/2026

One of my favourite quotes from the piece

010

Reposted by Ben Hall

Dan Davies @dsquareddigest.bsky.social · 10/09/2026

It contributes to the impression of dodginess that most of the time it's "LLMs are totally inscrutable, we can't tell anything about the relationship between training data and output, we can't explain any decisions" and then when it becomes inconvenient they're "nope, definitely didn't happen"

14612