I'm an electrician.

I dare you to use ChatGPT to wire a plug.

This is why AI is absolute horse shit.

#chatgpt #aibullshit #aiisdangerous

@nomenloony i hadn't zoomed in or read the post or alt text and am a *very* unsophisticated user when it comes to anything electrical, and at first glance i thought the correct chart was on the left and the AI-generated pic was on the right

once i read the text body and inferred that the thing labelled so confidently "correctly wired" must be the slop, then read the alt text and confirmed it was so, everything made so much more sense

AI infographics are such an awful case of "we can lie faster than you can fact-check"

@nomenloony@nomenloony.com @grok which one of these is accurate?

@nomenloony Live to Earth - yikes! Even looks wrong to me and I only know basic electronics.

@nomenloony looks like Google's Gemini got it right.

https://g.co/gemini/share/c0bcb787ae74

@skyfire747 @nomenloony The problem with AI is you don’t know when it’s wrong.

@drahardja @skyfire747 @nomenloony @blogdiva

My favorite quote about LLMs is that they are extremely effective for people who already are domain experts on the topic

The UX fail of the century is for every one of these companies to not append “teach me how I would verify this is true?” to every single prompt.

The skill it takes to use the tech is higher than the avg user is going to invest. Every single maker of chatbots has ignored this from day one.

Social darwinism.

@nomenloony What the flippin’ hell 😳🙀
Just so ya know I am *NOT* an electrician and I am freaking out at this image 🙀💀!!!

“We’re doomed” 💀

(Please play in your head Bernie Sanders’ doppelgänger/distant cousin when he said that exact line 😹)

@nomenloony
My understanding of AI ( which is none really) is that it generates a solution, answer, whatever is being sought, from a vast source of data.
So then , is the vast source it collects it from, wrong to begin with. Or at least the majority is wrong, being non human the IT machine can't see that, so it assumes it's giving the correct answer or solution to the request and so the human who requested it believes it to be correct because it's IT.

I'll stick to human to human for now.

@BigD It generates some text, or in this case an image, that looks like an answer. The problem is, ”looking like an answer” is often not a very useful property, unless you’re trying to find something to write on a test, which is all these machines are trained to do.

@ahltorp Then it will have the correct answer in time to come I assume. Just needs to learn more. Sounds almost human -:)

@BigD it does not learn

it is trained by humans who think something is wrong to not repeat the exact same thing, and with enough iterations the result are things that to laypeople look alright

like the image on the left