this post was submitted on 02 Jun 2025
243 points (98.4% liked)
Fuck AI
3003 readers
712 users here now
"We did it, Patrick! We made a technological breakthrough!"
A place for all those who loathe AI to discuss things, post articles, and ridicule the AI hype. Proud supporter of working people. And proud booer of SXSW 2024.
founded 1 year ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
He didn't cite wrong information (only) because of ChatGPT, but because he lacks the instinct (or training, or knowledge) to verify the first result he either sees or likes.
If he had googled for the information and his first click was an article that was giving him the same false information, he would've probably insisted just the same.
LLMs sure make this worse, as much more information coming out of them is wrong, but the root cause is the same it's been before their prevalence. Coincidentally it's the reason misinformation campaigns work so well and are so easy.
Edit: removed distraction
Chatgpt is unfortunately fully capable of generating false information without ever being given it.
What? LOL then why are so many companies suing them for copyright?
Because it's complicated. It is fed that data, but it can't access it, refer to it, look it up or anything like that. If you feed it all of reddit, you can't just ask it what comments did this user make, it simply doesn't know. It uses all the data it's fed to build statistical patterns of language and concepts, which is what it then outputs.
This is why it can quote things like Shakespeare, because that information is so widely repeated, it's fed it so many times, it's a common pattern it can reliably reproduce. But it isn't looking in some database and finding that Shakespeare quote to repeat, it doesn't have that ability or information.
But it does look it up. It even links to it.
If you have it search the internet yes, that's competely different to its default behaviour though. That's specifically providing it a document to look at after it has been trained, which it can look at and refer to.
If you're looking up content written by humans and published to the internet in an article, it is far less likely to be wrong.
It's a bit less likely to be wrong, but there's plenty of room for it to be wrong, either maliciously with intent or through incompetence of researching even basic things on their part. Someone being wrong once by misreading, or without interpreting data, or by trying to steer perception of something, can easily snowball into many sources concerning that wrong information ("I've read it, so must be true"). Many kinds of information are also very dependant on perspective, adding nuance beyond "correct" and "false".
There are plenty of reasons to double check information (seemingly) written by humans, it's just good to double check that for different reasons than ai content. But the basic idea of "it can easily be wrong" is the same.
No. Far less likely.