All my comics have happy endings!

man. i hate it when mastodon people are right about things. that's annoying af

Wait, Tyler Cowen has been writing about how to find virgins since 2009 while he was a professor at George Mason University

TODAY IS THE ONLY DAY YOU CAN REPOST THIS!!!

Someone should probably inform the White House's "AI & Crypto Czar" that no one is forcing AI companies to train their models on Wikipedia

The anti-digital ID petition is rapidly approaching 2.5 million signatures petition.parliament.uk/petitions/73...

Just gonna put this here. No reason at all. en.wikipedia.org/wiki/List_of...

Incredily ballsy for the government to include India's Aadhaar as one of the inspirations for the UK's #DigitalID programme. If I was trying to convince people of the benefits, I'd want slightly less focus on a system that lost private data about a billion people. en.wikipedia.org/wiki/Aadhaar

Wait, the new line is it'll help poor people? Will a digital ID reduce rents? Build council housing? Fix homelessness? Make the benefits system easier to navigate? Treat chronic health issues? Improve employment rights? It's marginally less shit as an argument than fixing potholes but not by much.

Cannot stress enough how it is already possible to report potholes www.independent.co.uk/news/uk/poli...

turns out chatbots are shitty search engines too tommorris.org/posts/2025/i... by @tommorris.org

New blog post! I've previously tested consumer genAI tools with legal questions. Result: lots of popcorn. Now I tried a professional genAI tool. It's less bad—it doesn't cite fake cases. Instead it cites irrelevant cases and doesn't really offer much benefit over just using a search engine.

Medical researchers proudly confess to straight up data fraud

🤷🏿‍♂️

We could remediate the effects of a bad process with bad technology, or we could abolish the bad process.

If you think AI “makes a phone feel smart”, I can only conclude your definition of smart includes not being able to work out which states have the letter “r” in their name.

On the same day in the Telegraph:

This rules. Google's AI search summaries answers questions using material from the Hypothetical Hurricanes Wiki, a community of people that write Wikipedia-style articles about made-up hurricanes. (Also, in the universe where Hurricane Humberto actually happened, Biden is still U.S. president.)

I'm remembering a time when there was a "collective delusion about the allocation of trillions in combined capital". Sure, it led to a global financial crisis. But we got a decent movie out of it starring Christian Bale.

Just need to implement an is_reasonable function then we can get rid of the entire pesky justice system. Now, let’s see, how do we implement this function?

just checking in with one of the most influential philosophers alive today

I thought vibe litigation was a high risk activity, but turning to a chatbot for medical advice is an even worse life choice. arstechnica.com/health/2025/...

AI services for $1 you say? Time to dust this off... Written during the pandemic (you can guess which company was in my mind) but still applicable. www.jenitennison.com/2020/05/19/p...

This giant wooden horse they’ve rolled up outside our city looks really cool. And they want to sell access to it for only a dollar. What could go wrong? We need to consider how far we’d fall behind if we didn’t embrace the advantages that the giant horse might give us.

New post about how we regulate The Algorithm (you know, the one we're always worried about), and one of the problems with the #OnlineSafetyAct. Namely, picking out the complex recommendation algorithms used by big platforms like TikTok and YouTube is hard. tommorris.org/posts/2025/w...

One of my all-time favorite exchanges on Github is Douglas Crockford telling one of the bootstrap authors "learn to use semicolons properly" and the author responding, "i have learned to use them, [sic] that's why there isn't one present"

Let's see how Bluesky's supposed decentralisation works out.

I never understood the appeal of Substack. When it started, I muted the term on Xitter to avoid people banging on about it. Blogs have existed for decades and you don't have to deal with an arbitrary corporate landlord.

You basically cannot.

Hail, Dorian.

Right, it's been A TIME, and it might be too late, but we have a fantastic website thanks to @natbuckley.bsky.social. Anti-SXSW is vaguely go! We just need you to help make it happen. If you'd like to run an FKFJ event, or get involved: fkfj.london

great job Gavin

Correction: the first North American pope…

German legal closed-access publisher Beck raises huge amounts of money to train proprietary German LLM and calls it ‘sovereign’. This isn’t ‘Eurostack’, this is a new step in the enclosure of access to legal knowledge. techcrunch.com/2025/04/22/n...

First they came for the queers, and you didn't know it because we're not in the poem

Worth noting that a number of universities have now sued over withheld and canceled grants, but no university has yet sued over the arrest, detention, and threatened deportation of its foreign students. www.nytimes.com/2025/04/19/o...

Thanks to AI, schools can focus on their core mission: being a daycare facility for teenagers where nobody ever learns anything.

I wrote a thing. It's long and rambling and concerns the important topic of the vibes around tech. It has spoilers for Adolescence and s7e1 of Black Mirror. And I spend way too many words getting infuriated about Roblox. If that hasn't put you off... tommorris.org/posts/2025/t...

My first chat with the Meta AI on WhatsApp. Anyone surprised that neither option works? (Android WhatsApp app, likely the most common way of all.)

I think Meta's going to win the AI race lads!

This is a super interesting example of everything that goes wrong with tech policymaking. Many hundreds of hours of good people's time spent trying to define a nebulous concept with a good name but no defined use case.

This is my entire life

Reminded today of the most effective and pragmatic approach ever devised for enforcing word count limits.

I asked DeepSeek R1 to summarise a government document and turn it into "an easy read summary aimed at those with cognitive, learning or neurological disabilities/differences" as a test. The prospect of having to parse British legalese seems to have induced an existential crisis.

The free speech crowd will be on to this, right? www.lbc.co.uk/world-news/d...

Somehow it always seems to come down to "blame people for being neurodivergent", rather than "fund a system which provides decent diagnosis and support". Btw, a lot of that money goes into helping neurodivergent individuals function in this world, and therefore able to work. Without it more can't.

In a decade, perhaps everyone will be capable of producing more inanity than the most inane politicians can today.

Oh no, what if China outperformed the US at synthetic text generation? How would we cope? Our mediocre business email summariser robot is actually of key geostrategic importance, I'll have you know. arstechnica.com/tech-policy/...

Caught up with the detail of Starmer's speech, and there is a real dissonance between the getting closer to people and communities and cutting bureaucracy *inserting more AI* into government. Deploying AI in this way requires *more* rules and processes, not less www.gov.uk/government/s...