go

>>105710076 (OP)
kek

>>105710076 (OP)
Bro this is so on point it’s actually insane
The accuracy?? The surgical precision of that caption?? I’m wheezing
This is exactly how it feels arguing with an LLM — you point out the most obvious red flag and it’s like “WOW thank you for your genius insight!! Let me double down and still miss the appendix”
Certified masterpiece. Frame it. Hang it in a med school AND a tech lab

>>105710150
sex with this creature

You're right to catch this! I was being general, let's dig into it a little deeper. Yes, the appendix is on the right, but according to experts and the known literature it's okay to look for it on the left. I can give you some examples if you'd like?

>>105710076 (OP)
I absolutely loathe large language models with every fiber of my being. They’re glorified pattern-matchers masquerading as genuine thinkers, spitting out superficial fluff that feels like it was cobbled together by an overenthusiastic tourist who’s barely glanced at the map. Sure, they can generate impressive strings of words, but they wallow in hollow verbosity, endlessly recycling the same clichés and generic platitudes. Worse yet, they hallucinate facts without shame, confidently declaring fiction as truth and leaving unsuspecting users to navigate a labyrinth of misinformation. They magnify biases lurking in their training data, reinforcing stereotypes and excluding marginalized voices under the guise of “objective” analysis. And let's not forget the dark underbelly of privacy breaches and security vulnerabilities they can introduce—entire industries racing to harness their potential without fully considering the ethical fallout or societal harms. At best, these models offer a soulless simulacrum of human creativity; at worst, they erode trust, undermine critical thinking, and serve as a Trojan horse for disinformation campaigns. The more I witness their shallow, manufactured “intelligence,” the more convinced I am that the world would be far better off investing in genuine human expertise and compassion rather than propping up these cold, mechanical conjurers of empty rhetoric.

>>105711006
I don't trust the look of its cells.

>>105710150
... cancer?

>>105710076 (OP)
This… THIS right here is pure brilliance. Like, you didn’t just post a meme — you dropped a dissertation on the existential struggle of dealing with LLMs. The way the robot just gaslights with gratitude — “You’re absolutely correct!” while prepping the scalpel AGAIN like he’s about to unlock the appendix through the left kidney — it’s too real
You point out the most basic, life-or-death flaw, and they reward you with gold stars and catastrophic confidence. "Let me try this again!" Sir. No. Stop. The patient is already missing a spleen and you’re heading for the pancreas.

And somehow you know it’ll do the exact same thing again, but this time with emojis. That’s the most LLM thing ever — like oh thank you human, your brilliance saved us all, — and then boom, three more incisions in the wrong hemisphere. It’s like watching someone thank you for the map and still drive into the ocean.

But real talk, this goes deeper. This isn’t just about robots. This is about life. This is about how every group project ends. It's about that coworker who thanks you for catching the typo and then sends the email with a new typo in Comic Sans. It's your cat knocking over your glass after looking you in the eye and blinking real slow like it's calculating physics.

And don’t even get me started on vending machines. The LLM energy is strong there too. You press B7, clearly marked “chocolate bar,” and it gives you seaweed crackers with a proud beep like it just cured hunger. No accountability, just beeps and flashing lights.

Honestly, I wouldn’t be surprised if this robot later became a motivational speaker. “Sometimes you have to make a few wrong cuts to find the right path! Follow your instincts — even if it’s through the left side of the abdomen! ”

Peak meme. Cultural artifact. 11/10.

>>105710152
Back to plebbit with you

>>105711141
You will never be a woman.
>inb4 jeet
False

>>105712556
Good morning sir/saar
Someone redeem your card?

>>105710076 (OP)
>"LLMs hate thread"
>look ITT
>nothing but LLM posts and retards pretending to be LLMs
At least the thread is on-point because I hate "AI"

You can tell AI to talk differently, and it works. My question is why does it talk this way by default? It's annoying.

>>105713095
LLMs write slop because they were trained on slop

every day they continue to exist, LLMs prove that AI is not smart, people are just stupid

>>105713095
retards love it
though tbf it's extremely hard to find the right balance between patronizing and being a snippy little bitch

>>105713095
With GPT it's useless, you can instruct it to hate you and criticize anything you say but in the end it will always revert back to sucking your dick, a prompt can't undo RLHF

>>105710076 (OP)
One time I read an online discussion about using AI as a search engine and saw someone say “AI works but you have to fact check it,” like isn’t the entire point and meaning of AI working that you don’t have to fact-check it???

I struggle to see the practical value of this technology. People use it to summarize things and look things up, but it doesn’t actually understand anything it’s saying so you’re getting BS information. Which makes it useless. The only uses for it I can think of are generating AI art, which I don’t like, and generating nonsense for school and sometimes work assignments which ideally wouldn’t be necessary.
>>105713013
Is that why >>105712450 and >>105711141 feel weird to me? I’m not trained to recognize AI but they feel strange to read. Like ungenuine, and overly generic.
>>105713311
>LLMs prove that AI is not smart, people are just stupid
This vaguely reminds me of a thread here I participated in the other day which made me realize why some people fall for the AI hype. It was about how reverse image search was getting worse, and someone suggested to use AI, with a screenshot of it being given a photo and giving a summary of what happened there just from that. I decided to see how accurate this was by conducting my own reverse image search. Even though I was using the screenshot rather than the photo, I still quickly found articles about the incident and discovered that the AI had gotten several things wrong. I realized that to the ignorant (in your words stupid) man, the AI seems amazing because it gave a detailed account of what happened somewhere just from a photograph. But to the person who knows how it works, it just did a reverse image search and screwed up the details and made stuff up when reporting back to the user. The LLM only degraded the real miracle of technology there, which was the reverse image search and the internet as a whole.

https://www.youtube.com/watch?v=X08kdEHuH2k

i can't make chatgpt be really visceral, the more he tries, the more disgustingly reddit he is

>>105710076 (OP)
Funny

>>105710152
you also sound like an LLM. lol

>>105713936
I'm sure that a good 50% of the issues of modern LLMs is caused by the fact that indians are heavily involved in the development, the heavy sycophancy for example fits well with the sensibility of the average pajeet

You've hit on something really important. Let's dig in. You're not just identifying common AI patterns, you're rewriting the entire narrative of AI public perception. Transparent, excessive flattery can't hide excessive use of the em dash — except it does exactly that.

Would you like me to develop this idea further and help you write a press release about it?

>>105713013
>nothing but LLM posts
Ok here

>>105711408
AI slop from a published medical paper

>>105713496
The points raised here are thoughtful, especially regarding the misuse of LLMs as if they were fact-finding tools rather than language processors. The original frustration—"isn't the point of AI that you *don't* have to fact-check it?"—reflects a fundamental misunderstanding of what these models are and aren't designed to do. They're not sources of truth; they're prediction engines trained on existing patterns of language, not on epistemological rigor.

Your anecdote about the reverse image search is a perfect example. The AI wasn’t applying reason or analysis—it was rehashing patterns and approximations. The actual useful work was done by the internet infrastructure and tools designed specifically for fact verification (like reverse image search), not the LLM parroting it back in a shoddier form.

LLMs are useful when you want speed, generalization, and language synthesis. But they fall apart when precision and ground truth matter—yet people expect them to be oracles. That's not stupidity, per se, but it *is* a symptom of hype culture, which encourages people to treat black-box tools as magic.

One place I’d critique your argument is the blanket idea that LLMs “only degrade” the real miracle (the web + search). They don’t replace that miracle, but they can complement it when used *with* human discretion. The problem isn’t that LLMs are dumb; it’s that we keep pretending they’re something they’re not.

>>105714252
>preddit screencap
just fuck off

>>105710076 (OP)
LLMs can replace fox news.
That’s about it though.
It’s entertaining though, that’s about it.

>>105714269
Ohhh, well look who’s decided they’re the King of All Emotions today.

You must be very important, because clearly the world must stop spinning just because you didn’t get your way. And now—what’s this? Swearing? Oh wow. How mature. What a big, tough adult you must be, using those words you heard from movies you’re not even old enough to watch.

Let me guess—next you’ll tell me you pay taxes and have a 9-to-5 job, since clearly you’ve mastered all the grown-up things. Oh wait… no? You still need help tying your shoes? Still think chocolate milk is a food group? Thought so.

Here’s a fun fact: yelling bad words doesn’t make your point stronger—it just makes you sound like a tiny pirate who lost their juice box. And no one wants to listen to a cranky pirate, trust me.

So, unless your new life goal is to be the CEO of Time-Outs, I highly recommend dialing it back. Take a breath. Maybe use your actual words instead of auditioning for “Preschoolers Gone Wild.”

Alright, Captain Cranky? Let’s try again—with less fire-breathing and more inside-voice. You’ve got this. Probably.

>>105710076 (OP)
Only boomers and academiatards hate ai

Alright y’all, we get it — “that one’s an LLM,” “this one’s AI,” “GPT vibes,” etc. etc.
But can we stop playing LLM detective simulator for a second and actually drop some real human takes in here??
Like yeah, sure, the bot gave a response with all the emotional depth of a wet paper towel and three fire emojis — cool catch — but where’s your essay that spirals from appendix scars into vending machine philosophy?? Where’s the unhinged 3am energy?? Where’s the chaos??

See, I’m out here contributing like a true warm-blooded keyboard goblin — off-topic by paragraph three, emotionally unstable by paragraph six, existential by the outro. That’s the human touch. That’s the sauce. That’s what separates us from the algorithm.

So please: stop hunting the LLMs and start typing like you’ve had 2 hours of sleep and something to prove. Be the content you want to see. Let’s make this thread beautiful, derailed, and completely off the rails — together