Thom Scott-Phillips

Some personal news: gulbenkian.pt/gias/news/th...

Scattered thoughts on how doing math feels these days dynomight.net/math/

A 60% increase is a disaster. And that is on top of 22% and 13% increases in previous years. Plainly AI contributes, so does mass precarity among early (and mid!) career researchers It's obvious that the model needs to change. ERC must show leadership

Working‑class academics are still struggling to fit in at university theconversation.com/working-clas...

When you stop to think about it, it's absurd. When I apply for a grant from, say, the local council to install solar panels, I don't get to include some % of the overheads I incur maintaining the house in the first place Grants should be for project costs. Not a way to service ongoing costs

This is Dennett's preface to Millikan's "Language, Thought & Other Biological Categories". Whether or not you agree with Dennett or Millikan, it's an excellent opening that both puts the material in perspective in a concise way

I have started using LLMs as tools to aid my own thinking: bouncing ideas around, etc. It's funny how different LLMs have different personalities. I find myself going to different ones for different purposes, even within the same project

Really happy to share the news of our interdisciplinary research group on human an artificial conversation. Exciting stuff coming up! #ai #philsci #sts #literature #rhetoric www.the-new-conversation.com

A Little Time The Beautiful South www.youtube.com/watch?v=2lGz...

Seems that nobody has attempted to use Millikan's notions of direct and derived functions to characterise or understand the notion of a construction, from construction grammar. I've found brief asides to this idea in a few places, but no detailed analysis. This gap is slightly surprising to me

Gun control has something in common with philosophy of language

“The dubious achievement of prediction markets is to have found a new way to exacerbate and exploit the simultaneously frantic and passive experience of the world that is a defining aspect of our age” www.lrb.co.uk/the-paper/v4...

Gun control has something in common with philosophy of language

Very interesting on how the current moment is completely transformative for mathematics Imv, there is a potentially very bright future to be had but institutional l/ship has to think hard about the incentives. Career rewards and culture must re-directed towards high-quality understanding

Very interesting on how the current moment is completely transformative for mathematics Imv, there is a potentially very bright future to be had but institutional l/ship has to think hard about the incentives. Career rewards and culture must re-directed towards high-quality understanding

This is fantastic satire "In keeping with our commitment to objectivity, no human has watched any of the endings. Internal research found that human viewers introduce bias, emotional responses, and the persistent question "wait, who is that guy?"" www.illposed.claims/posts/openai...

Wow. Looking just at the thread and the abstract, this seems like a genuine novelty and major advance. Will have to read deeper

Small but regular income to as many people as possible -> good research.

This is an excellent initiative. Historically, psychology has not much valued descriptive research, and this has been a mistake. Good to see a leading journal pointing the other way www.psychologicalscience.org/publications...

The greatest trick the devil ever pulled was convincing the world to use agentic language to describe a new technology

I asked Google what would be in the readme file

I asked Google what would be in the readme file

This is a future of student assessment, imv The essay will be (say) 20% of the mark, and it'll be fine/expected that students use LLMs The other 80% will be a short oral exam defending the thesis in the paper

As common-sensical as this might seem on the surface, I think it's mistaken and I am sad that research l/ship does not seem able/willing to question it Short thread

"*The teacher swam the boy" is pretty grammatically unacceptable. But it's much more acceptable if she gets into the pool and moves his arms and legs, rather than just blowing her whistle and shouting. Why? - Sentence constructions have meanings in and of themselves: lingbuzz.net/lingbuzz/010... 1/n

Interesting to consider with this relevance-theoretic approach to fine art....

"You are cheating people of the authenticity of their moment" Best discussion yet of the Manchester City verdict and what it tells us about modern football

Why is Nature trashing its brand? I've noticed all sorts of daft stuff appearing in its pages the past couple of years. Not so in Science

Back of the envelope calculations: 3600 years of research time went into the collective application process. The research will fund 3200 years of research. We are 400 research years poorer as a result of this. 1/2

i've read from many that "gold standard science" sounds good in principle and that could not be farther from my truth. there is no such thing, there cannot be such a thing, and the sheer thought that it could actually be a good idea betrays a complete ignorance about science and scholarship.

This is a future of student assessment, imv The essay will be (say) 20% of the mark, and it'll be fine/expected that students use LLMs The other 80% will be a short oral exam defending the thesis in the paper

I’m in Portugal, and I am finding it quite odd to be in mainland Europe but on the same timezone as the UK

I'm tired of how agentic language is used to describe AI models. This language avoids placing responsibility where it should be, with humans When a plane crashes we don't think *the plane itself* is somehow morally responsible. It's safety regimes, pilots, etc. Same here

High valuations are sustained by trust; trust is built on accountabulity; but AI companies won't hold themselves accountable as.ft.com/r/b63376a8-1...

Yes this. I don't like it when the infrastructure makes it wiser for me, as a cyclist, to go on the pavement, but you can be damn sure that when I do I go slow and give pedestrians lots of space