Do you believe him?

I’m reminded of the bit that the Nigerian prince email scammers purposefully include typos to select for the people that would fall for the scams.

That sounds incredibly normal to me though. Both that approximately no one would either commission art nor have an original character, or actually put a value on this art in the first place.

ahh! This looks incredible. Would love to run this on the Grafana codebase 👀

WHY THE MR BLOBBY JUMPSCARE?!?

Think of all the farriers that are out of jobs because we stopped using horses to get everywhere.

idk, I don’t really find this that compelling. Is a company using a camera to take a photo immoral because they took a job away from a portraitist?

I can’t shake the notion that Pluribus is an allegory for queerness and gay conversion ‘therapy’ (a phrase a despise). Ever since she brought up a few episodes back I can’t stop thinking about it. I think it really applies both ways. Who is Carol to think she knows best?

Imagine if UI was actually functional, like an actual interface for users.

How can you, with a straight face, call this better? Does this even contain a text box?

This was made by someone who hates computers and despises this functionality existing at all.

yeah, apparently! According to... the data that comes from my smart meter (if im applying it correctly)

Honestly my smart home setup could use some more integration testing

Products like Cursor and Claude Code are doing wonders with agentic AI because all the agents were basically already built for them. It's honestly magic watching them iterate away.

I can see how tempting it is for execs to see this and generalise it out to everything, but we're still leaps away.

Then there's the recognition that large *language* models are really good at writing programming *languages* because its the exact thing it's trained for.

Trying to shoehorn non-language goals into an LLM will only ever be as your ability to express the domain in pure text.

Even before AI, there are entire subfields of computer science dedicating to *proving* that the code is Correct. We assume the human "hallucinated" the code so we'll run a bevvy of automated tests to make sure it doesn't break things.

Your light switches don't assume you'll turn on the wrong light.

@reckless.bsky.social @davidpierce.xyz I believe the core reason for the disconnect through all the AI hype is that programming is uniquely suited for AI in a way that's hard to generalise to other fields because it already assumed the human made a mistake.

It does not warrant being a 6 hour film. If that's how we should judge it, then its even worse.