Some sense checking: ONS life expectancy projections for the UK compared with observed life expectancy (black line). Six years of likely over-projection compared with decades of under-projection

updated fable eval scores

Similarly, I tripped the Fable guardrail by asking about dragons: jonminton.github.io/jon-blog…

One hour trying out Claude Fable, the new Mythos class model, for a data visual task requiring the kind of aesthetic judgement earlier models usually struggled with. Very impressive! jonminton.github.io/jon-blog… jonminton.github.io/claude-f…

I'm on a mission to make this plot as publicly recognizable as the one with the holes in the WW2 airplane

The single most important DataCamp course on Agentic AI isn’t about AI: jonminton.github.io/jon-blog…

How will your country's demographics change this century? At Our World in Data we built a tool to let you see for yourself. The UN's assumptions are the starting point. But then you can adjust the three drivers of change — births, deaths, and migration — to what you expect.

New blog: digital dogs in hats: bridging the senseworld/streamworld divide: jonminton.github.io/jon-blog…

My cat rendered the way I think he sees himself. In this case a dormant firegod.

The fact the y axis is per capita whereas the pessimistic scenario is human extinction makes this even funnier for stats pedants

Vaguely reminiscent of Simpson’s Paradox

Musk’s Evil Detector is set off by being an AI guru who used to be a game developer who once made a game called ‘Evil Genius’

It’s already happened. (And it’s a bit underwhelming)

Dear @RichardDawkins, you've always been an inspiration to me. I made this website for you. My goal is for it to help you understand AI chatbots at a deeper level, and avoid getting fooled by sycophancy and other cheap tricks that models have learned through RLHF. dearricharddawkins.com

This is how LLMs with extended reasoning manage to one shot correct answers: youtu.be/-HCCwTJ7PVI?is=Ap4O…

I've always been kind of annoyed I wasn't better at Rubik's cube solving. So a Gemma 4 design session while on a plane w/ no internet a Claude Code ~one shot yielded this Rubik's solver/trainer. Create your cube or paste a flat-net image of your scramble → it reads the colors, finds a 22-or-fewer-move solution, and animates the 3D cube through every turn. No backend. 100% in the browser. Run it: jeffhuber.github.io/rubiks-s… Code: github.com/jeffhuber/rubiks-…

Two and a half episodes into Margo’s Got Money Troubles. It’s not yet turned into a loose gender swapped remake of Breaking Bad but that’s the hope.

We need to keep reminding ourselves of what's been happening to patient flow in Scotland's acute hospitals over the last few years - particularly since the summer of 2021.

Any sufficiently advanced technology is indistinguishable from magic that can be magically fixed by turning off and on again

An increasingly coherent picture of the impact of AI on jobs, by @jburnmurdoch @ft: 1. New Fed paper by Crane and Soto now confirms with official labor force survey data what private payroll analysis was showing: roughly 500,000 fewer coders are working than pre-LLM trends would predict. 2. Argues evidence consistent with my work (with Lin and Wu, link in my pinned post) on weak/strong bundles: junior developers and contractors hold "weak bundles" (their work is mostly standalone coding that AI can substitute directly), senior developers hold "tight bundles" where coding is combined with domain expertise, judgment, and cross-functional responsibilities, making substitution much harder. 3. Freund & Mann and Gans & Goldfarb add a second lens: what matters is the value of the tasks that survive automation. Remove coding from a senior role and you free up time for higher-value work; remove it from a junior role and almost nothing remains. ft.com/content/b69f8599-eaf1…

I read this too quickly, and wondered why Campbell was claiming Rayner is a ‘money dwarf’!