I think people would take the vow of silence in the quiet carriage more seriously if they were also required to wear a monk’s cape and a tonsure wig.

Imagine you time travel back to 1847 and you find the left response to industrialization is a) machines will never be as good as human weavers or b) we need to copyright loom patterns or c) it’s a speculative bubble. You’d say “Y’all. Not helping. What you need is obviously a labor movement.”

I cannot believe there are still people in my mentions arguing that boycotting an election because the candidates aren’t addressing your issue is a way to get them to pay attention.

Anti-intellectualism in the most literal sense: pro-dumbness.

The "overdue" timer in todo list apps indicates how long it was since you gave up on that attempt at organising your tasks.

In fact, I'm beginning to suspect it thinks I'm in a closet instead of a bubble.

I'm curious about the Discover feed's algorithm. It shows me content very similar to what I usually read, and a sprinkle of things I'm extremely unlikely to be interested in, with little in between.

This morning my ChatGPT quota was inexplicably exhausted. It took a while but I pieced it together. Voice mode somehow got activated when I went to bed. The bot then engaged in a 10 hour conversation with my snoring dog, answering questions the pup wasn’t asking and praising him for his insight.

At its core, hate is intellectual laziness.

It would be funny if we eventually develop therapies based on the naked mole rat that radically extends lifespan but makes you look like Nosferatu.

Remember, if you are going out in the hail showers today, pop a child's xylophone on your head for added fun.

No, I do not want to swipe sideways just because the first two pixels of finger movement were horizontal.

Fortunately for them, toddlers didn't achieve alignment of the superiorly intelligent adults.

YouTube should understand that when they show me ads for snake-oil that explicitly dissuade people from seeking necessary professional care, it reflects very badly on them too.

I instructed Gemini to be a very senior developer, and it told me it's time to call it a week... 😐

Much like how AI is said to be useless yet somehow also a threat to people's jobs.

It's important to remember that Trump has used more than just rhetoric to foment political violence. His entire 2nd term has been a theater of cruelty in which violence is portrayed and practiced as the ultimate measure of political seriousness and resolve.

These threads about how humans are superior to AI because we know when we don't know something are always funny, because they're full of people demonstrating the opposite, including @mcuban.bsky.social himself when he confidently makes the dubious assertion that we will "always" have that advantage.

A skeptical mindset isn't about distrusting others. At the core, it's about challenging your own beliefs. Scrutiny of others just follows as a natural consequence.

This is an interesting case study, but I suspect the focus on whether or not they should shift right obscures a deeper issue of a general lack of initiative and narrative ownership from Labour. Being seen as simply mimicking your opposition and lacking authenticity is always going to harm support.

There are four possibilities for artificial super intelligence: 1. It can't be controlled but benevolent 2. It can't be controlled and is evil or indifferent 3. It is controlled by benevolent people 4. It is controlled by evil or indifferent people Risk of 4 isn't getting as much attention as 2.

I find it curious that it's not the soft and tender, modern bread that's the woke one, and instead the traditional, rustic one.

A common subcategory of vibe coding is Dunning-Kruger vibe coding.

I'm contemplating what would happen if elections were replaced by randomly selecting members of the public for political positions, as a duty, and think there's a decent chance things would actually be better.

Every coding agent I've tried so far is like a very fast, highly educated and attentive developer with terrible instincts.

...potentially comprising other models, specialising in other cognitive functions, like memory, motor control, etc.

my most pessimistic anti-luddite spin is "there is no way off the bus and you should probably worry more about who's driving it"

I think these policies are getting increasingly problematic, even with the opt-outs, considering how much usage around self-help, coaching, and similar has been growing recently. People no longer use these services just for novelty.

This is why "but the democracy!" is an ineffective campaign massage. Politicians have to sell something people already understand the value of, or explain the value.

Seems I accidentally exfiltrated some of the GPT-5 system prompt. I'm using an "introspection step" to debug my bot, where it's asked to analyse the combined set of prompts it sees, and point out any conflicts or confusions using explicit quotes. It decided to quote the system prompt too. 🙂

And here I have been telling people that what I like most about GPT-5 is that it's more boring.

The first one-person unicorn won't be due to an LLM doing all the coding. It'll be due to an LLM doing all the menial paperwork so the person can focus on coding.

They say you get dumber by having ChatGPT think for you, but reliably catching where it's bullshitting you on some advanced subject requires Sherlock Holmes level awareness and brain activity!

Refusing to "trust the experts" is meant to be a brave, responsible position. But (as Carl explains in the thread) there is too much knowledge in the world. You have to trust *somebody*, much of the time. If you have no trust in experts, you end up trusting the idiots.

I expect we'll see a new paradigm of software emerge. Currently we have LLMs writing the logic for static apps, and we have static app utilising LLMs to process data. The former is inflexible, the latter unpredictable. Apps where LLMs dynamically update the logic based on data could solve both.

This is why I sprinkle trick questions into my ChatGPT dialogues. 😅

My latest cartoon for @newscientist.com

I'm curious what will happen when Trump understands that US companies pay license fees for foreign IP on medicine, tech, etc. If the attention goes to some specific instance, I'm not convinced it will settle things that the US is (presumably?) a net recipient of cross-border payments in total.

As LLMs get progressively better at coding, will we see a re-emergence of live, reflective environments like Smalltalk was, where the application code isn't separated from the interface? Why should I not be able to tell my app to add a feature, if the app itself knows how to code?

lol Python is so useless This programming thing is clearly a bubble.

GPT 5 is much more stubborn when it's wrong, oh my...

I wrote about what went wrong with ChatGPT's sharing dialog, and why I think it's reasonable for people to be confused by what looks at first glance like a very clear checkbox description simonwillison.net/2025/Aug/3/p...

Aside from the ideological trend towards privatisation and outsourcing in recent decades, I feel there's this self-reinforcing cycle where public services operate poorly, lowering public perception in them, which then makes it less attractive for talented and driven individuals to work for them.

The biggest driver of robotics innovation over the last couple of years hasn't been anything clever. It's just been people making cheap robots easier to use. And that's a good thing; we have all the tools we need for a robotics revolution, we just need to actually get people out there using robots.

It's smaller than bacteria. Almost like a molecule, but mean. — My 4yo, trying to recall the word "poison"

Hear me out: A robot vacuum that detects and collects Lego pieces in a separate compartment.

It's funny how similar this is to some career advice I've read.

The April 1st mental escape route is golden!

The criticism against LLMs that they were "built to replace people" (debatable) is disingenuous even if taken at face value. Every single tool ever made was built to "replace" human labour or allow us to do new things. Not to mention the whole industrial revolution, that's enabled modern society.