edit: fixed thumbnail
this post was submitted on 02 Feb 2024
677 points (97.6% liked)
Programmer Humor
32730 readers
108 users here now
Post funny things about programming here! (Or just rant about your favourite programming language.)
Rules:
- Posts must be relevant to programming, programmers, or computer science.
- No NSFW content.
- Jokes must be in good taste. No hate speech, bigotry, etc.
founded 5 years ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
Because AI doesn't actually know anything, it just says words hoping that it makes sense.
Well… it’s a correct phone number. So that kind of undercuts your message.
edit: I’m actually a bit baffled by people downvoting this. That is the correct number given by both of those organizations. It isn’t some LLM hallucination.
thats like saying theyre wrong because words are spelled correctly yes the number is correct but the machine doesnt know what the hell it is, or what it's for, or in any sense "understand" what it's regurgitating to the user as evidenced by the fact that it listed it twice. "AI" doesn't know anything, it just copy-pastes shit.
First, it just copy pastes much in the same way animals do; a neural network with outputs weighted by experience. Secondly it posted it twice because both of those organizations are real and are references for the topic it mistakenly meant to reply about. The same way of asking what to do when a house burns one might reply:
Third, and most importantly, I’m not saying it invalidates the message completely… but it does undercut it. As in, there would have been a much stronger case for just randomly outputting garbage information that it hopes sounds correct if the information had not been, you know… correct.
meanwhile i asked it to write a short simple hello world in a scripting language designed for children, and it spat out nothing but garbage. one of us is leaning on confirmation bias.
I’m curious which language and which model, because I have had several of the models write programs like the sieve of Eratosthenes quite successfully. You can find this report in my GitHub of the same name.
I don’t know what bias you’re on about. I was just reporting that those phone numbers are in fact the correct numbers given by those organizations. Are you implying they aren’t? Because, you might want to go to the primary source and check for yourself.
you're saying that AI has hopes