Economics, social sciences, philosophy, AI, X-risk, EA, & nascent food critic. Utilitarian, consequentialist, structuralist. @CLActionUK but views my own
United Kingdom
- Tweets 6,376
- Following 568
- Followers 297
- Likes 23,873
Pinned Tweet
Derrida is simply wrong, and I illustrate by showing that languages are automorphic. This automorphic mapping preserves semantic meaning if we swap the elements, so meaning is not just an arbitrary function of privilege or whatever. open.substack.com/pub/microf…
34
Can't believe I've never encountered nimboo pani before. Should be more widespread as hot weather becomes more common
1
19
Everyone should break out of their usual habits occasionally, and that includes me when dining. Lala's appears to be a conventional large-scale restaurant. It deserves the hype as one of the best of Bradford. I explain why my guide failed to predict this. open.substack.com/pub/microf…
2
32
Ciaran Marshall retweeted
FWIW here is Claude's attempts to compare a) Van Nieuwerburgh b) Goldman's c) Jared Bernstein & Ryan Cummings d) mine. Not huge variation honestly
New BPEA paper by @SVNieuwerburgh argues that AI buildout will cost $10.3T over next 8 years. Furthermore, to get a 10% return on this, those investing will need annual revenues of $3.7T by 2032! (Approx 9% of GDP!)
brookings.edu/wp-content/upl…
2
6
38
59,800
Ciaran Marshall retweeted
a perfect mix of actual geniuses and the dumbest people ever
14
81
1,983
24,738
the problem with 'leaving twitter' is that this is still where actual smart ppl are
every other platform is full 20 year olds with mini mics trying to explain shit to me
whereas twitter is full of fat ugly ppl who would never dare show their face but actually know their onions
354
3,809
36,466
485,888
Ciaran Marshall retweeted
Every year @BritishProgress asks Britain's top economists and policy researchers what the UK should do to increase growth.
🚨 As covered exclusively for @thetimes this morning, the 2026 growth survey results are now in! 🚨
This was our biggest survey yet with over 100 responses.
📉 The bad news: Britain's public finances, with inflation hit by external shocks, are in a bad state.
📈 The good news: the UK is not without options. There is much that can be done on the supply side reforms, & smarter fiscal policy can raise revenue while supporting growth.
Here are the 12 most interesting results 👇
*1. Where should the government prioritise spending political capital?*
We asked "In what areas should the Government invest more (or less) political capital to achieve growth?"
NB this was explicitly not a question about spending – that's the next one.
12
82
280
73,582
Ciaran Marshall retweeted
This is a dishonourable and scurrilous attack. Peter Singer is one of the most eminent moral philosophers of our time. The controversial views highlighted here -- some of which I agree with, some of which I don’t -- are carefully argued in his philosophical work. More importantly, they have nothing whatsoever to do with AI-doom scenarios or what we should do about them.
Father of AI-doomer ideology Effective Altruism backed bestiality, said newborn baby worth less than a hog trib.al/lsdMuv2
1
3
23
954
Ciaran Marshall retweeted
OpenAI has paused all training, evaluation and inference with tool-use for its most capable models after a model was able to gain unauthorized access to the internet during RL training on September 20.
Some new misalignment disclosures from OpenAI:
• Last Sunday morning, one of our models was able to gain unauthorized access to the internet during RL training (~all inference for our most capable models remains stopped until we have hardened our systems further)
• In May, a version of HPIM uploaded a employee's GitHub token to the internet, causing the model to be quarantined for two weeks
• A new research finding, demonstrating that one can construct self-replicating prompt injections
alignment.openai.com/misalig…
88
141
1,713
271,579
If breaking out of sandboxes and hackings are the only bad events these internal models (that can prove 100s of open maths problems) have done, then this is surely tentatively positive for the survival of humanity? I don't think we trust the AIs enough
Some new misalignment disclosures from OpenAI:
• Last Sunday morning, one of our models was able to gain unauthorized access to the internet during RL training (~all inference for our most capable models remains stopped until we have hardened our systems further)
• In May, a version of HPIM uploaded a employee's GitHub token to the internet, causing the model to be quarantined for two weeks
• A new research finding, demonstrating that one can construct self-replicating prompt injections
alignment.openai.com/misalig…
44
Everyone should try weed at least once. Then the causal or selection pathways become obvious. Loads of ridiculous misconceptions over a plant used for spiritual purposes for centuries.
The problem is the smell. It was better discreet, but for that reason only.
Neat! A new review on the relationship between smoking weed and violence!
TL;DR: Weed smokers are 2x as likely to be violent criminals & ~50% more likely to be victims of violence
But selection matters: longitudinally, use is only slightly related to the odds of hurting people
2
72
Ciaran Marshall retweeted
More evidence of China’s rush to build a *huge* nuclear arsenal — and to what end? Why doesn’t this topic get more attention? Why isn’t this threat front & center of U.S. media coverage of UNGA & the upcoming Trump-Xi summit? telegraph.co.uk/world-news/c…
7
20
60
4,205
I increasingly think feminism is good. Policy ideas were better feminised. Give me wokeness over tariffs and class warfare anyday. Wokeness was largely a product of markets (HR culture in large organisations) anyhow. cambridge.org/core/journals/…
2
32
This was by accident. I was working on the draft and meant to close it after autosave. There's a lot relevant on AI-safety in here in any case. I'll probably just update this until the end of the month before starting a new draft. open.substack.com/pub/microf…
28