In the sticks…
In the wilds of Donegal. What estate agents, those masters of euphemism, would call “a development opportunity”.
Quote of the Day
”I think that maybe if women and children were in charge we would get somewhere.”
- James Thurber
Musical alternative to the morning’s radio news
Mark Knopfler | Ahead Of The Game
Long Read of the Day
Trump’s 5-Step Fascist Plan
Robert Reich in sombre mood. His question: how could Trump actually turn America into a fascist state?
His answer: In five steps, which he’s already signalled he intends to take.
The steps, in a nutshell:
- Use threats of violence to gain power.
- Consolidate power after taking office.
- Demonize a group of people and establish a police state to round them up into detention camps.
- Jail the opposition.
- Undermine the free press.
If you haven’t been following what’s going on in the US, you probably think this is scare-mongering. If so, maybe you should have a look at Project 2025: Building now for a conservative victory through policy, personnel, and training. It’s a detailed plan for a comprehensive takeover.
When Trump came to power in 2016, he hadn’t expected to win, and so had no plan for governing. Hence the chaos of his first administration. If he wins this year, though, he would come into office with a plan. And, as Robert Reich points out on his blog (and in this video, it all looks remarkably like a plan that someone else adopted in Germany in the mid- to late-1930s.
In his remarkable book, How Democracy Ends, my friend David Runciman argued that democracies never fail backwards but forwards (i.e. in some new way). The implication was that looking for models in Europe’s decline into fascism is misguided. I’m beginning to wonder if that was too complacent a judgement.
Books, etc.
This arrived today. I’ve been looking forward to it. Neil is the DeepMind Professor of Machine Learning in the Cambridge Computer Lab and one of the most thoughtful experts on ‘AI’ I know. It comes out in the UK on June 6.
Chart of the Day
‘MMLU’ stands for Massive Multitask Language Understanding. It’s defined as
”a new benchmark designed to measure knowledge acquired during pretraining by evaluating models exclusively in zero-shot and few-shot settings. This makes the benchmark more challenging and more similar to how we evaluate humans. The benchmark covers 57 subjects across STEM, the humanities, the social sciences, and more. It ranges in difficulty from an elementary level to an advanced professional level, and it tests both world knowledge and problem solving ability. Subjects range from traditional areas, such as mathematics and history, to more specialized areas like law and ethics. The granularity and breadth of the subjects makes the benchmark ideal for identifying a model’s blind spots”.
What the chart suggests is that powerful LLMs are proliferating — and that CO2 emissions and water consumption are increasing proportionately.
Linkblog
Something I noticed, while drinking from the Internet firehose.
- Willie Nelson and Kermit the Frog sing Rainbow Connection!
This Blog is also available as an email three days a week. If you think that might suit you better, why not subscribe? One email on Mondays, Wednesdays and Fridays delivered to your inbox at 6am UK time. It’s free, and you can always unsubscribe if you conclude your inbox is full enough already!