@DJDarren @winkleink @cstross On the contrary. I think he should be the *only* person required to use these instructions to wire up a plug.

@th @cstross This made me quickly move my hands away from the phone. It still makes me uncomfortable just looking at it. Excellent!

@DJDarren
LOL looking at the picture with WTF on my face, and after while read that it's AI genera^W hallucinated.

@winkleink @cstross

@shaknais @oddhack @cstross

I have nothing but:

@cstross Oddly, you might live. That drawing was SO %$&*ed up that it appeared to have a whole extra pin!

@th @cstross

No, just no. It's a good quality socket (MK are one of the best, and used to be the automatic choice for the building trade), but you have no idea what's behind that faceplate and if it's been correctly wired. Sticking _anything_ conductive into an electrical outlet is a very bad idea. (And a metal door key is probably one of the worst things you could use.) 3:O(((>

@Phosphenes @josh0 @winkleink @cstross In fairness, there aren't many sockets that could accept that plug.

@cstross ah the elusive 3-phase UK plug

@dotcolm @cstross for countries that use 0VAC

@DJDarren @Phosphenes @winkleink @cstross you know what they say: when you have a hammer, you can solve any problem with it.

@elaine1helen @cstross Yes, one key point of failure for asking "simple" questions of genAI is that they are designed to generate something original, combining and randomizing their training inputs instead of just copying one.

And for something like this, there's the issue that they don't really have a 3D model of the world, so every different angle of an image is like a different variation on the diagram, and these then get combined in ways that don't make sense from a real-world perspective.

@th @cstross Yep that's fine. Everyone knows sockets use Yale keys, but it was a Chubb you'd be in trouble

@cstross #3 often applies all on its own, independent of the first two.

@cstross

Imagine the bomb disposal trope. Countdown running out. Hero has to decide which wire to cut, asks a LLM
BIG

Wiring a plug was part of the Standard Grade Science course in Scotland (going back a few years). Towards the end of its lifetime we were forbidden from teaching it on grounds that you are no longer able to do it. It did not come up in the exam.

@ehproque @cstross

Ow my soul

@SteveClough @cstross

We were expected to learn how to wire a plug at school (late 80's, probably). My dad had already taught me, so I got full marks for doing it properly -- short live length, some slack in earth, and so on.

I then demonstrated to the teacher that I could do it again (properly, still) with a butter knife and my teeth...

@cstross unlike the first image this one isn’t plausible enough to follow

The first image will mislead people

@cstross gotta watch out for that BLIVE pin.

@ErikJonker
But that is still quite different from the large language model-approach which is used for the 'foundational models ' like Claude and ChatGPT. The LLM approach has structural weaknesses and is the cause of the enthusiastically pushed falsehoods. As it does no know anything and has no real world feedback, every output is always based
on probabilities deduced from its corpus of inputs and training.
In specific cases, within specific/limited areas of competemce AI has proven very useful. But, that is very far removed from AGI. So far the LLM approach is mostly misused for internet slop, useless reports and articles and avoidance of responsibility by governments, like Israel's or ICE target selection...
@sotolf @cstross

@pst @sotolf @cstross NotebookLM has proven it's value, at least for me. AGI is a different discussion which I don't find very interesting. Just "strong AI" will change our world.