Ryan J. Yoder

Last Notes

2026-10-26 19:35:54 UTC

- reply

I'm curious which way you expected fedi to lean...

2026-08-06 22:46:31 UTC

- reply

🤔 I wanted to say the combustibles themselves but the more I think about it, I'm leaving towards it's just fire. It's sorta like saying what's a tornado made of. It's not just air. The fire includes the light and heat it's giving off. It's more of a phenomenon than a physical thing.

2026-07-30 03:00:24 UTC

- reply

Would you add your own alt text as a comment?

2026-03-15 22:08:28 UTC

- reply

I think that LLMs will be seen more as a new auI for interacting with computers, not general AI.

2025-10-06 21:22:39 UTC

- reply

Is this like a 3D printed PDF?

2025-09-02 18:21:28 UTC

- reply

Is this a confession or a complaint?
Also how is cash more expensive...?

2025-07-27 13:55:02 UTC

- reply

I don't think the statement is that controversial. Yes we understand how to train a model but we don't understand their interworkings. This is just one example:
https://deepmind.google/discover/blog/gemma-scope-helping-the-safety-community-shed-light-on-the-inner-workings-of-language-models/

2025-07-27 13:16:43 UTC

- reply

I get that. But I think it's an oversimplification. The truth is we don't really understand how these models work. Saying an LLM can't lie because it's just a statistical model, is a bit like saying humans can't lie because our brains are just a bunch of molecular interactions. The research into mechanistic interpretability is pretty interesting (not that I understand it all). The work shows that at a basic level specific concepts and ideas are being encoded in specific nodes in the model.

2025-07-27 12:38:08 UTC

- reply

If you didn't know the implementation of a specific model would there be any way to determine if it was capable of telling a lie?

2025-07-25 17:15:07 UTC

- reply

If an llm can't lie, then it seems true by only a technical definition. I can ask Meta ai basic questions about geography and get correct answers reliably. And I can ask it to lie about basic geography and I can reliably get incorrect answers.

2025-07-25 15:37:42 UTC

- reply

Maybe this is semantics but Meta AI appears to be able to lie.
https://files.techhub.social/media_attachments/files/114/914/603/628/133/633/original/40e1a232570b03a4.jpg

2025-07-12 12:06:31 UTC

- reply

Maybe this is just semantics, but the US is not "going to war over Taiwan", they would be defending Taiwan from a Chinese invasion.

2025-05-23 22:09:11 UTC

- reply

I personally haven't experienced this a lot but it sounds like something you just have to correct Junior devs not to do. And it will take time. Don't let a junior dev get away with pushing code they don't understand.
If senior devs did this, I'd be very irritated and I'd have a serious conversation with them about it.

2025-04-08 11:04:49 UTC

- reply

I'm posting this specifically because you mentioned screen readers. It took me a while to figure out what was going on but once I did I could see it would be hard to play the game. In any case, this is very cool technically. I'm curious to play the JavaScript version to see how usable it would be in comparison.
https://files.techhub.social/media_attachments/files/114/302/003/730/272/390/original/a2680b6f663125b3.mp4

2025-04-03 20:43:01 UTC

- reply

When the British mock us for using imperial units it feels like your group of friends all agreed to dye their hair blue, and then you show up at school and are the only one who followed through. Come on guys this was your idea!

2024-11-02 11:51:19 UTC

- reply

there are certainly grey areas, but I don't think these movements are in conflict with liberal democracy. Other forms of government certainly are in conflict, such as theocracies, single party states, military juntas. I think we generally agree and maybe this is mostly semantics.

2024-11-02 11:07:44 UTC

- reply

"8) Accepting that not every country is, or should be, a capitalist liberal democracy" - putting capitalism aside for a moment, which people don't deserve liberal democracy?