zoraniko
11487
276
3
Apr 10, 2026 10:45 PM
zoraniko
11487
276
3
ArcaneM37
And he's my ex-husband. Ugh
merrowcattech
I so agree. My roommate had gone so far over the edge as to name her AI. Drives me crazy with it.
EMHPicardo
jezebel2718
Fuck me that's terrifying. And all the narcissists out there...
discostewsm
Can't remember where I saw it, but a study has shown that these models are roughly 50% more agreeable than a typical human. So it's willing to give a wrong answer to please people. This is one reason why folks are declining in intellect.
OperatorWay
LLMorons. Third-party thinkers with second-hand thoughts.
Hmuda
"You're right, that was the wrong answer, thank you for being the smartest smart-boy in the world, only you could have spotted that error. Here's another, even more wrong answer."
bawtzki
Also has memory of a goldfish, contradicting itself from 5 answers ago
tinyfootprints
I'm wary of free AI. If using the software is free, then the user is the product.
williamfny
And that means the someone more unscrupulous than you or I could poison the data they are trying to collect... Shameful really if someone were to do that.
BerryButcher
chatgpt was OK before but now it's fucking trash, ever since they downgraded the free model to be a complete imbecile, then sold out to Department of War it became extremely biased towards any peoples rights questions, videos, i deleted my account few weeks ago, deekseek free model is leagues above this garbage despite being china limited one
Eidodk
ChatGPT has NEVER been "OK".... it is a LANGUAGE model, it is NOT trained on facts, it IS however trained on peoples questions, reactions and responses... If you ask ChatGPT itself if it is reliable, it says no based on the fact it often hallucinates answers and data that it ISN'T trained on.. That is - by the way - the EXACT same problem with each and every other AI chat bot.
BerryButcher
It has been better than google at one point, it looked up what was asked, it expained things accordingly, at one point it started making up information, then started loopbacking and backtracking on what it said, creating disinformation, became unreliable, chats could no longer be sustained because if did it would simply confuse itself, it was OK for info searching at one point, i think you're misunderstanding something, i never treated it as anything but search engine.
tinyfootprints
I asked ChatGPT to write some C++ code for me, and it did a decent job. For research, I already know how to use a search engine. I'm just not buying in to all the excitement over AI.
Hatlled
I use AI when programming my Pi Pico. I am AWFUL with programming, and it helps with problems- like the Raspberry Pi site giving up coding that simply doesn't work, etc.
I can make an LED light up, and I can *just barely* get the wi fi to work on the thing. :P
tinyfootprints
I'm not bad at programming. I could have written the code myself, but I really just wanted to see how well ChatGPT would perform. And it was fairly decent. I don't think I would trust it to write major parts of a program, but for the nuts-and-bolts type of code that runs dialogs or other user interface functions, ChatGPT can crank out code faster than I can, so it is useful as a time-saving tool for me.
loma45
Claude code is wildly impressive - and I say that as a software developer. It writes better code than most people in the industry, it just requires you to have enough understanding to keep it in line.
As long as you are making the design decisions, laying out the architecture, and keeping its tasks fairly focused it will write pretty damn solid stuff. Not flawless, but solid.
If you then have it review the code in a separate session, it'll catch a good chunk of the bugs it created too.
Wolfshead009
Problem is too many people who DON'T know the code expect to be able to tell AI to write code and have it work perfectly.
ongabonga
AgamemnonsMemes
Yeah we absolutely can stand to make stupid people more stupid. We're so cooked as a species.
EMHPicardo
BallsToYouMyGoodSir
never used any of them
donttellmywifeimup
https://time.com/7295195/ai-chatgpt-google-learning-school/
schmonday
To be fair, that just says that blindly accepting what ChatGPT gives you is what erodes thinking ability. Engaging with its output in a conversational way (ie. Thinking about its output critically) showed little difference from not using it at all. And that's true of anything. But, also, I do feel like people are more likely to blindly accept whatever they're told, like most people who just read this headline.
NZSheeps
But it's got diagrams!
ongabonga
It's not a headline. Read the paper. ChatGPT is dangerous.
schmonday
That's just a different headline. Again, engaging with it critically offers no danger as compared to any other method (as far as the brain goes).
MapleSyrupMafia
Just because a couple people went craycray because AI told them that their conspiracy theories were valid doesn't make it dangerous!! Yes, you can substitute floor with concrete powder, it will give texture to the cake! /s
Huor
But people accept what ChatGPT says blindly. That’s how it’s used.
schmonday
Not always, but you are repeating what I said.
Huor
I'm repeating what you said.