ACK!
AI is literally the most retarded thing. LLMs don’t think or reason. Literal zero intelligence. Anyone who falls for the AI hoax and uses it deserves to be lobotomized.

What does being correct have to do with being able to control people

>>512234120 (OP)
>he fell for the gAyI meme
Lol. Lmao. ROCLMFAO even

>>512234120 (OP)
>Obsessed with full moons
Are you a werewolf?

>>512234120 (OP)
Ask it about the jews and banks

>>512234120 (OP)
If you ask ancient models from 2024, you get these kinds of answers. Try this month's models.

>>512234120 (OP)
Maybe use an llm with reasoning and thinking turned on?

>>512234338
I love full moons. I love astronomy. Fuck that astrology bullshit though.

>>512234587
That was the newest grok 4 released last week. LLMs will never be intelligent. They can’t do basic reasoning. They don’t actually think. They don’t care about right or wrong.

>>512234597
You must be at least 18 years old to post here

>>512234120 (OP)
I wish people would stop treating AI as if it could do everything a normal computer can but also talk to you, like Data from Star Trek.
It doesn't do exact math. It recognizes patterns using statistics and uses that to generate stuff. That's it.
If you want accurate astronomical data, use a regular computer program.

>>512234355

>>512234120 (OP)
AI is spectacularly bad at questions about time. I don't understand why, but my guess is it's because people on the internet are bad at questions about time.
>>512234804
The newest Grok4 was released like an hour ago. It can turn memes into shitty 5 second videos in 15 minutes!

>>512234120 (OP)
>open chatgpt app i have installed for some reason
>ask that question
>get the correct answer
Whatever, expected that.
>since I’m there, ask it does it think, to see what it says
>it gives some bs response about it uses algorithms to draw conclusions and its not really human thinking but similar yada yada
>ask it again does it consider that thought or simple pattern recognition and response fetching
>feel stupid for sitting there and pretending to have a conversation with the computer and close the app before it even finishes typing
I still don’t get how people do it seriously, let alone get obsessed with it. I feel like a fucking retard using it, like I’m talking to an imaginary friend. Its like those carnival fortune teller animatronics, and thinking it was actually talking to you.
Like its actually embarrassing on a personal level for me.
Idk, I get I’m over thinking it but still

o3 got it right
Every other model failed

>>512234120 (OP)
now feed it articles from jewish newspapers, that‘ll make it really awesome and safe!

>>512236282
Ask it “when was the full moon in May, 1988”
I’ve noticed the LLMs fuck up on all the months with a blue moon (second full moon in a single month). May 2nd and May 31st for 1988

>>512234120 (OP)
>>512234804
You're still gonna get replaced. It doesn't matter how much you seethe, ai is gonna take over your job

>>512237055
>Hey AI, why is Poland Catholic?
>Poland is Catholic to escape the German cultural orbit of Protestants
>Poland, a country founded five hundred years before protestants, revolted against Protestantism by becoming Catholic
As a student of history, it is very troublesome that it cannot get the most basic things right. Or even close to right.

>>512234804
AIs don’t need to be smart. They just need to execute basic function of data imputation which is 99% of office drone jobs

>>512234804
You sure you weren't accidentally on Grok 3 homie? This is Grok 4

Just knowing a few basic facts (sidereal period, looking outside last night or tonight) you can work out any full moon in the past or future on a fucking napkin. If you haven't taken basic astronomy, you're missing out on timekeeping and how our planet functions.

Yeah I was trying to plan a camping trip with deepseek and chatgpt and they were giving me crazy places at crazy prices that didn't exist. One even gave me a script to say to a farmer and it was an old woman. The info may have been correct but none of the places really existed.

>>512234120 (OP)
woah

>>512234120 (OP)
AI will be useful when working with a large amount of information where there are many patterns. They already interpret medical tests better than living doctors.

LLMs are dumb in a vacuum. Their magic lies in their ability to generate coherent and varied strings of text from nothing, with loose generalist contextual awareness mechanisms made possible by extremely efficient and sophisticated embedding spaces, which was completely unprecedented only 5 years ago. Yes, LLMs are still bad at facts and sometimes reasoning, but fact finding and reasoning is actually very easily delegable to non-ML or specialized-ML tools like Google searches.
The majority of advanced in LLMs in the last 3 years in GPT-3 took off hasn't even been the AI itself, it's been the infrastructure around AI that allows models to be actionable and precise in deterministic ways. Especially with the widespread adoption of MCP these days, hooking up even the most rudimentary model capable of producing slightly coherent text, when given access to the agentic utilities that exist today, are infinitely more capable than even "next month bro" GPT-5 could ever be.