Fediverse

The beautiful toucan with the big beak. #myphoto

@dasgrueneblatt Die meisten Strecken, die ich fahre, sind recht angenehm und meist bin ich gleich schnell bzw. hin und wieder schneller dran als mit den Öffis 😁 Und notfalls kann ich mein Faltrad eh mit in die Öffis reinnehmen!

@Tutanota i misclicked 😭.
Yes ai, but only locally, and ethically sourced data. I find LLMs useful for translation, and small utilitarian applications of it.

#OpenJDK (!!!) has an interim policy on the use of generative "AI" until Oracle proposes a policy:

"Contributions in the OpenJDK Community must not include content generated, in part or in full, by large language models, diffusion models, or similar deep-learning systems. Content, in this context, includes but is not limited to source code, text, and images in OpenJDK Git repositories, GitHub pull requests, e-mail messages, wiki pages, and JBS issues."

https://openjdk.org/legal/ai

Many of us know about the Simple Sabotage Field Manual [1] and reference it either ironically or seriously.

Now. Here's the thing. I am 100% sure that it has found its way into LLM (Large Language Model) training data.

And that is why I think that "AI" is sabotaging us all ;)

#SarcasmButOnlyHalf #Oops

[1] https://en.wikipedia.org/wiki/Simple_Sabotage_Field_Manual

84 partners. 45 trade agreements.

Happy Friendship Day to our partners across the world's largest trade network.

🇪🇺 🤝 🇦🇱🇩🇿🇦🇩🇦🇬🇦🇷🇦🇲🇦🇿🇧🇸🇧🇧🇧🇿🇧🇦🇧🇼🇧🇷🇨🇲🇨🇦🇨🇱🇨🇴🇰🇲🇨🇷🇨🇮🇩🇲🇩🇴🇪🇨🇪🇬🇸🇻🇸🇿🇫🇴🇫🇯🇬🇪🇬🇭🇬🇩🇬🇹🇬🇾🇭🇳🇮🇸🇮🇱🇮🇶🇯🇲🇯🇵🇯🇴🇰🇿🇰🇪🇽🇰🇱🇧🇱🇸🇱🇮🇲🇬🇲🇺🇲🇽🇲🇩🇲🇪🇲🇦🇲🇿🇳🇦🇳🇿🇳🇮🇲🇰🇳🇴🇵🇸🇵🇦🇵🇬🇵🇾🇵🇪🇼🇸🇸🇲🇷🇸🇸🇨🇸🇬🇸🇧🇿🇦🇰🇷🇰🇳🇱🇨🇻🇨🇸🇷🇨🇭🇹🇹🇹🇳🇹🇷🇺🇦🇬🇧🇺🇾🇻🇳🇿🇼

Yes AI, but only locally - for example.

The public discussion on AI is still heated in 2026. What's your take?

Make your voice heard: Are you yes or no when it comes to AI?

If you choose, "Yes AI, but..." complete your thought in the comments.

of course, once you work out how to navigate the maze (which cannot solve your problem!) and speak to an actual HMRC person, they are decent and courteous and the issue gets solved in a couple of minutes. Nice one Dominic, thank you.

@EugeneMcParland slightly unrelated note, but it's great that you're no longer writing russian or russia with a capital letter

do the people who build HMRC's phone maze feel guilty about how much effort they put into making the phone maze route you to a "haha we hung up without you ever talking to a person" end point, or do they feel happy about having done it?

https://github.com/floreabogdan/birdy

What does my dog think I’m doing when I use the computer all day?

(Please do not reply to tell me that we should ban AI contributions entirely. I agree that it would be simpler than saying "it's fine but don't fix easy stuff with a robot" but for reasons we cannot.)

@canvas thank you everyone who, while working on their design, decided to keep my warm message towards the far-right Russian bootlicker AUR and their little dog, PSD. Much appreciated!

Our project has apparently ended up on some list of repos that agents try to fix easy tasks in. The problem is: our game deliberately has a large backlog of easy unfixed issues, for human learners to cut their teeth on. It often takes me longer to write the issues than it would to fix them myself, which is OK when there is an educational outcome, but now I waste my life reading LLM-written patches where no human learning has occurred. Time to add an AGENTS.md asking the robots not to do this...