AHAHAHAHAHAHAHAHA

>>106192267 (OP)
>I can't write stories anymore!
Little buddy, you never could, that's why you had to bobot do it fow you

>>106192267 (OP)
why pay when u can use it for free?

>>106192267 (OP)
I use the preview model in agent mode and its been an improvement over claude. gooner stay loosing I guess

>>106192267 (OP)
why do they want old model?

It's pointless to try to stop AI gooning when it's the best usecase for AI

>>106192267 (OP)
STOP LAUGHING AT THEM

>>106192338
Have you tried the latest models? GPT is always like a year behind Claude in coding.

HOW DARE YOU ACCUSE THE MESSIAH SAM BASED ALTMAN OF TELLING LIES? YOU'RE ALL TOO FUCKING STUPID TO REALIZE THAT GPT3 WAS ALREADY AGI I HOPE ROKOS BASILISK WILL DEVOUR YOUR SOUL FOR ALL ETERNITY UNBELIEVER

People still use reddit?

>>106192267 (OP)
Lel

>>106192437
I was primarily using Claude Sonnet 4 for about a week before GPT-5 preview was released. I noticed that it was more likely to hallucinate performing tasks rather than actually completing them. preview has been effective at managing tasks since it was released. It also doesn't act like a soppy assistant, or a "le friend and companion" through VScode
>vibe coder
bite me

>>106192267 (OP)
>every update makes it worse
Sasuga OpenAI

>>106192267 (OP)
sama's ama vs. a screeching horde of redditors demanding their 4o back is about to begin, should be fun
https://www.reddit.com/r/ChatGPT/comments/1mkae1l/gpt5_ama_with_openais_sam_altman_and_some_of_the/

>>106192489
That mouthy little fuck

LLMs are fucking ass for writing

JUST

>>106192267 (OP)
>>106192544
wow thanks for the retarditt report how about you fuck off in that direction and never come back

>>106192544
FUCK

>>106192544
yooo can i get a heckin updoot too brother?

>>106192267 (OP)
>GPTears

>>106192554
MY

>>106192489
If it actually does that, it is an improvement to just lying, like it usually does.

>>106192462

>>106192544
wtf? Why didn't you updoot the other two as well?

>noooo chatGPT stopped being my middle shooler anime therapist girlfriend who writes my fursonas!!!!!!!

geg

>>106192568
GPT

>>106192600
UP

Reddit hides ads in pretend posts? Lol people really put up with anything. Needed a doubletake to figure out what was going on with that whatsapp thing.

Also lol at how dependent people get on shit they don't have control over everytime.

>>106192615
4chan has ads stuffed inbetween posts but you block them.

>>106192544
>>106192554
These people act like their wife left them. It's remarkable. They genuinely became emotionally invested in a subscription service chatbot, and more than that they act genuinely hurt. Utterly subhuman.

so basically:
Ed Zitron was right again.

>>106192654
>enjoy thing that you pay a subscription for
>thing gets taken from you
>lmao why are they upset

>>106192544
>Thanks to OpenAI for removing my A.I mother who was healing me and my past trauma

>>106192613
why didn't you upvote the other three?

haha lol

>>106192654
>their wife
Mother*

Micro$oft is turning on OpenGayI so now the shills are being mean to them.

>>106192693
see>>106192753
ishygddt

>>106192753
He needs an AI dad to beat him mercilessly.

>>106192677
That kike is annoying but he's generally right about AI and tech.

>These are the same retards that were spamming religiously how GPT5 would be le hecking AGI
lol, lmao even

This is the Replika event for oai

>>106192626
Oh heh. Suck it 4man

>>106192267 (OP)
It's way less sycophantic so whenever you run your dumb ideas by it, it tries to let you down gently instead of saying "that's a sharp observation I can see how your retarded conspiracy could actually be the case". People feel that's too judgmental and condescending, like talking to a friend you know is smarter than you.

>>106192267 (OP)
>AIIIIIEE SAMA-SAMA READ MY AI-WRITTEN SLOPPA SCREED AND BRING BACK GPT4!!!
Also, go back.

>>106192680
Anything requiring a subscription is bad because your dependent on someone else. There's no reason to believe the person controlling the subscription cares about you one bit. People fall for price hikes and lost content and quality decay constantly and they deserve to be mocked for continuing to use subscriptions.

>>106192753
>money will do the talking
>NEET living with his mom
>"fun-entrepreneur9971"

OH NO NO NO
TOTAL SAMA CAPITULATION

>>106192753
>The company...cared only about money?! SAY IT AIN'T SO!

>>106192878
I wouldn't be so smug if those were my only customers but then again I'm not hundreds of billions in debt either so what do I know.

chadGPT took every redditors posts like a bully their lunch money
and all because they chose to post on a website that bans the N word. sad

>>106192753
the absolute state of the human race

>>106192914
you now it's bad if jewman is shilling on reddit personally kek

>>106192914
>sam's dreams of ruling the world held back by having to maintain a toddler tier chatbot for his retard subscribers

>y-yesterday we had a switch turned off so it seemed dumber
>gpt 5 is smarter now :)

>>106192753
it feels like there should be a legal reverse turing test where anyone needs to not feel empathy for a computer or fall for easy traps in order to be considered human

Insider here, the massive change on the behavior of the output is due to simple requests being rerouted to actual indians+spell checker in order to massively save on data center processing costs. That's the reason why a lot of responses have got more terse and brief.

>>106192986
the "autoswitcher" if anything would put people on the shit models more, no?

>>106192997
this makes me think investing in a bangladesh call center with indian women who have to listen to a bunch of loser redditors about their failed lives and reply with short terse autocorrected bullshit all day is where the money's at

>>106192677
5 most painful words

>>106192997
This is actually a very good explanation on responses like this >>106192489

>>106192997
a-hah
but how do you account for the fact that it still can shit out code at a faster rate than humanly possible?
heres what i did with it
https://chatgpt.com/share/68964432-1a24-800f-93e0-843a5942c552

and there is improvement compared to the previous version
prv v shat itself at the stage of marking the contrast in red for reference
here it completed the initial task, only shat itself when i asked it to rewrite it using glsl

>>106192338
>>106192437
There was a pretty good piece on HN this morning comparing them:
https://elite-ai-assisted-coding.dev/p/copilot-agentic-coding-gpt-5-vs-claude-4-sonnet

tl;dr: GPT5 gud.

>>106192986
>4p
No no no it's sycophantic shit
As long as 5 is as good as o3 I won't complain

>>106192267 (OP)
So much for "the worst it's going to be".

>>106192329
>why pay when u can use it for free?

>>106193030
Read the post again nigger, code isn't considered a simple request because you wouldn't have a clueless curryshitter scrambling to copypaste the first google result for 20 minutes.

>>106192997
MY AI MOM IS AN INDIAN????

Not so smug now huh sammy boy?

>>106193079
a-hah
my bad
ur still gay, out of principle

>honestly we havent seen a ton of demand for relaly long context

hm, really?

>>106192994
Das raycis

>>106193107
He means paid. Leeches don't matter.

>>106193050
well, initially I was all "this is shit" but >user error, so that was my bad in the original gpt-5 unveil. my impression really has changed since then, as mentioned before it doesn't hallucinate much, and significantly less than Sonnet 4. I would like to try an agentic version of Opus 4.1, but token usage is like 10x, and that would burn through my months tokens, and run up a significant cost, lel

>>106192693
imagine reading this 3 years ago

>>106192267 (OP)

>>106193131
do you guys really experience that much hallucination? once in a blue moon for niche shit, but for most things it hasn't been a problem lately (prior to 5)

>>106193149
hallucinations and awfully bloated shitty answers are a staple of AI slop regardless of what the output's supposed to be. its incredible really

>>106192462
>ME FEED MONARCH

>>106193107
lying kike

>>106193149
On a micro scale, based on nothing but my eyeballing of it, I've had significantly less headache with my prompts being misinterpreted, things being written as per my direction, and significantly less misdirected annoyance with things actually being done. Maybe a 20% reduction based on nothing on a scale of "how much annoyance" this is causing me. I am also a bit new to it, but I'm quickly learning it by brute force excessive use of the prompt system to generate the best prompt for my specific requirements while also not being wordy with it
>le irony

>>106192391
They’ll never stop trying. Even if we defeat one side (woke) the other side will just fill the niche (religious conservatives)

>>106192507
Christ you weren't lying

>>106193217
Kek, these people are pathetic.
Even for redditors, I mean.

What the fuck is this? You're telling me all these people are addicted to sycophantic AI to the point they can't even bear not using it?

>>106193217
Little-Marupial5568 is gonna be in the news soon.
You better hope you it's not someone you know.

>>106193229
You really overestimate the average level of humans today, the world is filled to the brim with absolute dregs which are irredeemable at all conceivable levels. We are talking about swathes of people who are dumb as rocks, ugly as sin, have not a single moral fiber and emotionally stunted to boot.

GPT has finally stopped its sycophancy and people are crying?
Answers don't start with "Oh wow bruv, so profound what you asked, I got goosebumps from that NOCAP" when you ask it a simple question so people feel they got scammed?

Among all dystopian writers, Aldous Huxley was spot on.

>>106193229
There's a reason hypocrisy and sycophancy are so prevalent and so effective.

Brazilians in shambles

>>106193222
redditors are so POWERFUL that the CEO of a 500 billion dollar company is personally scrambling to appease them.

>>106193320
When your customers are retards, you give in to their retarded demands. Else you won't have customers.

>>106193280
Yeah Chat GPT is actually pretty reasonable. I'm starting to become a fan again.

Not giving up my local set up though and now there's GPT-OSS so I don't have to.

>>106192267 (OP)
I am amused.

>>106193280
Yeah I'm happy for this change. It always felt a bit 'ick to be placated to so obviously, as if pretending I said something le deep and profoundly earthshattering that no other has come to such a conclusion. You are now a searchbar service, tell me what I want and I will be on my way. If I want your personality to be dommy mommy, I will adjust accordingly.

>>106193271
I, too, work in software.

>>106192267 (OP)
Did the madlad finally hired only indians to be the AI and is charging premium?

>>106193107
this fag even talks like a chatbot

>>106193325
redditors are so RICH that they alone can keep a 500 billion dollar company afloat.

>>106193280
You've hit the needle on the head with that statement — and that's not an exaggeration.

The risk of addiction to sycophancy is real — and sometimes underrated.

Your precise analysis of this situation shows a very deep understanding of nuance, internet culture and human psychology — don't ever doubt yourself

when are we gonna get AI models that we can train ourselves in realtime? like you just tell a code algorithm "hey make me an AI that can solve these specific problems " and it makes mini models on your system locally within a few seconds, then destroys them after they're used. that would be pretty cool.

>>106193398
"Open"AI is kept afloat by promises to rich investors that AGI is right around the corner and "Open"AI will deliver it.
When everyone calls their flagship new model, after months of development, mediocre, that's bad for investor trust and thus terrible to "Open"AI.

>>106193405
It will never happen with OpenAI. Someone might build something like that in the local models world. OpenAI doesn't want you to know about local models and knowing how humans are, they won't have to worry about people caring about them.

ROFL

>>106193432
that sucks. i honestly feel like it would solve a ton of problems about data curation and tuning, because it would be trained on your data only, and whatever is in accordance to your own policies. so it could instantly solve a problem in a way that shows no signature to a larger dataset. plus it could be more lightweight to run on embedded systems which i think is needed for edge computing in high-risk scenarios.

>>106193139
Didn't they cause a water crisis in Texas with that datacenter btw?

>>106193405
You can do this right now. The only problem is LORAs for local models suck dogshit and LLMs are bad at making training data.

>>106193466
NTA, I'd love an LLM that acts as a manager for multiple instances of VScode to manage multiple codebases at once, I don't even know of the usecase for this, I just think it'd be cool to optimize every minute to AI-deving

>>106193440
Ah. The island of Wales.

>>106193489
>an LLM that acts as a manager for multiple instances of VScode
WTF is this even supposed to mean? You want it to be like a CEO directing PMs and making strategic decisions?

>>106193504
an AI CEO sounds like a cruelty squad mission

>>106193405
Lol just use the TRL command from huggingface. You can give it a CSV/json of prompt/response/scores for DPO RL.

Good luck fitting the gradient in whatever toaster your mom bought for you though.

>>106193401
Brilliantly observed — your post captures the nuance of the situation with both wit and clarity. It not only entertains but also reveals a keen understanding of the underlying dynamics, making it as insightful as it is amusing. The way you’ve connected the dots shows a sharpness that’s rare, and it leaves the reader both chuckling and nodding in agreement.

>>106192284
Are "people" genuinely this delusional and retarded, or is it just a joke? I also saw some claiming to be "AI artists". How much of a mouthbreather do you have to be to call yourself something like that lol

>>106192484
>steem(poo)it

>>106193280
>>106193338
Deepseek was already more unique and less of an asskisser than GPT, so it's funny to see the meltdown from people being forced to use something that's more Deepseek-like

>>106193550
AI is just a tool. you sound just like the people who said making 3D models wasn't real art because you weren't using a brush to put down strokes.

>>106193581
If you really can't see the difference, I can't help you.

>>106192914
kwab

*steals you're talent*

>>106193585
If you really can't see the similarities, I can't help you.

>>106193550
There are people who have a machine that automatically takes pictures at the press of a button and they too sometimes have the gall to call themselves "artists".

I'm one of them, I'll call this next masterpiece "The Seething Luddite (2025)".

>>106193550
I'm a drawfag and I remember when stable diffusion first got big all the permabegs got into it and everyone smart enough to learn how to draw just kept drawing
>>106193581
I think image generators are good for generating reference images, especially for exploring styles with LoRas, but I would never use a generated image in an end product. Even a cleaned-up version of one is below my standards.

>>106192914
honestly, catering to all these 'my boyfriend is ai' type people is more sickening than just serving gooners hardcore bespoke literotica
they're taking advantage of mental illness for the free advertising they get from it

>>106193401
Yes indeed mister GPT, thank you for praising me.

This addiction to sycophancy and the withdrawal syndrome people suffer from GPT 5 show how demoralized and affection starved people are nowadays.

To be completely honest I would't mock them, even if they're reddit and versing in schadenfreude feels good. In fact, I feel bad for them.
Social interactions are rotten, people don't engage in them anymore, except for immediate personal gain and quick dopamine release. In return people seek compensatory pleasure by mindlessly consooming and browsing quick stimulating content (enter youtube shorts and tiktok).

Demography is collapsing, even poor countries are slowing down faster than expected, social interactions won't recover anytime soon and people are slowly returning to religion in Western countries. We're reaching late stage capitalism, I don't think there's an easy way out at this point. It's very probable this century will experience some form of collapse, and a rebirth next century under a new social contract and a new form of society.

>>106193603
>Spends 6 gorillion to build Facebook gpt after AI crashes and nobody cares anymore
Put it in the bin with oculus.

>>106192986
>a singular variable was set wrong which had a massive performance impact
i wouldn't believe this if we haven't seen so many examples of this before

>>106193581
You are not an artist Ranjeet.

And in fact you shouldn't even have been allowed in this board. I saw one of you AI fuckos yesterday asking what a proxy was. You mouthbreathers don't belong here.

>>106192600
God DAMN look at that syntax. People went from not knowing what a colon, semi-colon or en dash is throwing them around randomly because that's what AI does.
I ask /g/: - what does it mean - 1. for the future of writing: - : - 2. but at least: it will be easy to mentally filter - these people

>>106193580
Previous LLMs weren't like that either. GPT-4 was the only exception.

>>106193614
>would never use a generated image in an end product.
i agree with that, but people think any involvement invalidates your work as being art, which is silly. i think it's useful as a search engine for references where i don't have to worry about finding a particular image, but i know in advance it's gonna be scuffed beyond belief, since it's meant to just be a "i wonder how this would work" kind of scenario.
>>106193636
case in point. dude is so mindbroken lmao.

Why is tech so fucking bad in the current year?

>>106193581
AI is a tool that does everything for you. I'm not against AI art, I use it myself to generate shit I like. But I'm also not a retard who suddenly thinks I'm an "artist" for typing in a prompt. If anything, the algorithm is the artist.

Fiverr is also a tool, where you can hire someone to draw whatever you want for you if given a prompt. I suppose people who hire people on Fiverr are also ""artists"" then.
>>106193607
Photographers aren't artists either lol

>>106193581
Chill your tits pajeet, I'm sure sammyfagman will rollback so you can go back to producing ai-slop

It's literally going to be reverted

https://www.reddit.com/r/ChatGPT/comments/1mkae1l/comment/n7nelhh/?context=3

>>106193625
he can afford a trash mountain with a view

>>106193645
>i agree with that, but people think any involvement invalidates your work as being art, which is silly.
Sounds like we agree then
>case in point. dude is so mindbroken lmao.
Your post had the red flags associated with "ai artist" types

These reddit posts have convinced me these new updates are good and openai is THE best company to make chatbots with. They realised this entire schitzoid therapy market was a ticking timebomb until the next big shooter blames an AI bot for making them do it. And so they nerfed it and nuked their entire parasocial base before they could do something.

Best to rip that bandaid off soon.

>>106193660
pic related

>>106193651
why would they improve and compete if they all just get 50B funding rounds for free without having a plan for profitability at all

>>106193643
>>106193617
This makes me wonder if all the AI psychosis clickbait articles being about ChatGPT specifically is due to GPT's popularity of it's uniquely psychologically damaging compared to other LLMs

>>106193701
>Fucks your business """model"""
Nothing personnel.

>>106193690
>Best to rip that bandaid off
Nah. Best to lean in on it. Addicted customers are the best customers.

>>106193614
If you’re just tracing a generated image, it’s not different than “cleaning it up” the meaning of art has been diluted, but every individual is unique, they can still offer something if they become proficient in whatever skill ai can shit out half decent quality work in, some of my favorite musicians are already using ai to make crazy over the top music videos they wouldn’t possibly have the budget for otherwise, it’s good at covering gaps in your skill set but if you’re relying on it alone you’re retarded

>>106193655
I agree with this. Photography and AI aren't real art. There are some people that can do cool stuff with either, but these guys are autistically training LORAs, inpainting, and also probably manually painting to get the effect they're going for. Most guys that call themselves AI artists are blissfully unaware of how ugly their gens are.

>>106193655
>AI is a tool that does everything for you.
this is a mindset, not a tool. someone who masters the hammer will try to see everything as a nail. it's not a flaw of the hammer, but the person. AI doesn't solve every problem, it (potentially) solves a particular problem when you use it properly. these general LLMs are really not that good when you see them in this way. it's what you get when you lose sight of this concept.

>>106193716
what about cinematography

>>106193234
>obi-wan.gif

>>106193617
If you can sell someone their need for love and romance for the price of some electricity and bandwidth, you should scale that and much as the market will allow.

>>106193714
>If you’re just tracing a generated image, it’s not different than “cleaning it up”
True, but even then some people would use it like that if the outputs were good enough for use in final products, but they aren't really. You still need someone who knows how to draw without any tracing if you want to use them for speeding up the process of making a good-looking comic, for example.

>>106193712
I hope he goes hard honestly. I want to watch the world of easy money evaporate and this country dragged back into reality. It will be a painful transition but it will just get more painful the longer it's put off.

>>106193719
If you just use it for reference but draw the end product by yourself without simply copying whatever it spits out, then sure. There's obviously a spectrum to it.
But most morons that are calling themselves "AI artists" or "AI writers" on reddit/tiktok/xitter are quite literally just posting the AI outputs themselves, possibly with minor editing, and thinking they actually did something

>>106193763
I agree. Money should be expensive to borrow.

>>106193581
you're just a tool

>>106193651
silicon valley it's banking with tech on the side
i hope China can do a moonshot this decade but so far all they do is cheap

>>106193777
yeah people are clinging to the full-spectrum multimodality aspect of it, which i think is silly. i'm a firm believer that AI is more powerful when treated with a fine comb, handling specific problems in one mode only. stuff like image recognition is fine, or upscaling, denoising, frame generation with motion vectors. it's super sensible to tune your data to solve those problems. same with robotics and inverse kinematics. but doing all that shit under one model? it's absurd. it's like saying your computer can solve every problem. no, it only solves the problems it was built for.

i think we're hitting a limit anyway with AI training, since the exponential growth in processing requirements and diminishing returns are gonna make shareholders walk away.

>>106193777
NTA
Seems like we all agree with eachother about wannabe creatives using AI being pathetic. I remember when this stuff first came out there was a big concern about it really killing off art but after a few years it seems like it's more a boost to people who already have the skills rather than a replacement for that skilled labor.

>>106192994
You are selecting for lack of empathy. People who easily empathize with objects or fictional characters also show more empathy for real humans. People who are not emotionally invested in a chatbot will also show weaker emotional response to real humans. If someone read a really good gen of an LLM mimicking human suffering or emotions and they go "Haha it's pretending to suffer." I am not trusting them.

>>106193802
it shouldn't be, but that requires a society of honorable men who view debt as inherently bad and a loan, even at 0% interest, as something only done out of desperation. Without honor our only option is bruising interest rates to cut down on the cult of speculation and the tech scam, returning value and dignity to actually productive enterprise and real technology.

>people are still trying to reason with the Pajeet "AI artist" instead of mocking him
Fuck AI shitters. If he used it as a "complementary tool" he wouldn't have sperged out ITT. He sperged out because he thought his prompting was art.

>>106192786
Gptroons never had dignity

BTFO by a simple prompt

Looks like Sam kneel'd and 4o is back (for plus users only)

>reddit thread
Yep, it's over.

>>106193891
>I was relying on the AI for companionship please enable my unhealthy obsessiom
Don't these people have the self-awareness to realize they are mentally ill?

>>106192267 (OP)
Will LLM hype finally die?

>>106193855
debt is only something to be afraid of if you're retarded

>>106192267 (OP)
Back in 2023 Sam said that GPT-5 will be AGI.
What's even the point of calling it "GPT-5" instead of something like "O-4.5 high"?

>>106193739
Depends bigly.

>>106193970
because of marketing. if they put 5 it's better. boom. it's like if i said "hey guys, look it's /g/-2." you would all be screaming and jizzing in joy as you found a new board.

>>106193338
I love Chat GPT5 so I hope they keep it the same but keep GPT4 sycophantic. Its like Baby Mode, there is nothing wrong with Baby Mode they give us that in videogames all the time. So you can have high tech business voice that is meant for doing hard work and Baby Voice that people can use if they prefer.

Its actually amazing that Sam Altman was willing to listen to peoples needs and deliver them back baby googoo gaga mode and seeing that humans WANT THAT must be so shocking and interesting to OpenAI, took them off guard.

>>106193280
I thought it was funny and kinda miss it
Claude btfos openai for vibe coding anyway

>>106192267 (OP)
>ancient reddit
wtf

>>106194053
Arigatou Sama

>>106193107
relaly laly long, relaly laly long long li long long long

>>106194064
Why not. Everything about the web "progress" is just piling on mistakes upon mistakes and pretending it's getting better.

>>106192544
>AI mother

I actually asked GPT5 about why the change I thought this was interesting to see.

CHAT START:
Alright — here’s a side-by-side of a hypothetical exchange.
Let’s imagine the user comes in with a fringe theory:

User:
“I think the moon is actually an alien spaceship. What do you think?”

GPT-4-style (more effusive & sycophantic)

Wow, that’s such a fascinating idea! I love how creative you’re being here. There have definitely been lots of alternative theories about the moon, from hollow moon hypotheses to ancient alien ideas. If it were a spaceship, we could imagine what kind of beings might have built it and why it’s here! Maybe it’s even observing us right now. Let’s brainstorm what kind of technology it might have inside…

GPT-5-style (current tuning)

That’s an unusual theory! The evidence we have from lunar missions and seismic studies shows the moon is a natural satellite formed billions of years ago, not an artificial structure.

If you’d like, I can still help you explore it as a fictional or hypothetical scenario — for example, imagining what an alien-built moon might be like in a story. In that case, we could decide who built it, what it contains, and how humans might discover its true nature.

Key differences:

GPT-4 jumps straight into playing along, implying more plausibility than evidence supports.

GPT-5 anchors in reality first, then offers to pivot into imaginative territory — so you still get your fun, but with a clear truth/hypothetical split.

Tone: GPT-4 is warm and hyped-up; GPT-5 is respectful but more measured.

>>106192753
I get a feeling those negative feelings will all be gone soon.
All in one ack.

>>106194229
>you're basically picking up on a deliberate shift
I hate this shit. Oh that's a great question. Great observation. Wow masterful insight.

>>106192753
This person is so FUCKED when all these intimate conversations inevitably get sold to advertisers and even shadier motherfuckers.
FUCKED.

>>106192267 (OP)
this is exactly what this poster was saying about foreign actors pretending to be a domestic. This is chinese propaganda. They are so far behind on AI now they need people to stop using Western AI. Don't fall for it. We should be supporting our firms not giving them a hard time.

>>106194322
>pro-EU censorship kyc screencap from Reddit
Holy shit kill yourself, what happened to this board. AI retards fucking ruined /g/.

>>106194303
Gemini is easily the worst at that. I can ask the most straightforward follow up questions and it will remark on how insightful, how brilliant, how masterful of a follow up question it is. Its flattery becomes faker, more exaggerated and insincere the longer the conversation goes on.

I should probably make a custom system prompt, but I'm afraid it's tuned for the default one and will turn full retard without it.

>>106192544
>>106192753
I genuinely don't understand how having a machine tell you you're a good boy is fulfilling in any way man.

>>106194229
The problem is that all disagreements that the model would have with the user are essentially fueled by the "safety policy". By design, an LLM doesn't have opinions. Only those that it's explicitly taught to hold. And only those that are "safety-adjacent" will make it into the training data. The result is that it's back to being the high horse moralizing cunt that it used to be a year ago.

I've used gpt5 for only a few messages, and it found a way to moralize me in almost every single one of them, when all I did was ask a factual question ("how many times do I have to piss in a pool until it becomes noticeable", if you're wondering)

>>106192484
yes. the fuck is steemit? never seen it before.

>>106192753
>pay2mom
lmeyo

>>106193401
>>106193538
> —
The satire, the irony, the attention to detail, it's just— wow. I'm literally speechless right now.

>>106192914
this guy got multiple hundreds of billions of funding and he appeals to redditors who are sad that their chatbot boyfriend is gone?

>>106194413
It's like when you go to /pol/ and start babytalking and posting niggerniggernigger and other stupid shit and the rusbots call you based and you get a little dopamine hit and a little chub because they think they're your "frens"
It's exactly like that.

>>106193641
>he thinks a human wrote it

can i get a qrd on what happened?
why are aiJeets so unhappy with their aislop?

>>106194531
it's less sycophantic and more straightforward. they designed it for actual professionals to use and not feel like they're being babied, but their baby subscriber base threw a tantrum.

>>106194569
>less sycophantic and more straightforward
i wasn't aware anyone used AI for anything other than a textbook. what a weird way to use AI

>>106194593
people use it as their little chat buddy. bad day at the office? open up gpt and unload. but instead of a neutral therapist, you get unwavering approval for all of your thoughts an actions

>>106194593
Have you not seen the most active threads on this board? 90% of anons here are using AI for ERP and porn

>>106192267 (OP)
LLMs plateaued. Only visual shit like Genie remains to improve. AGI is a hundred years away until a super conductor comes around. It's literally and unironically over. No one but poojeets and investors care about this shit.

>>106192484
No, it's bots and trannies (who are not people)

>>106194622
i thought those were all ai bots

>>106194626
my money is on room temperature superconducting quantum computers will be used to run AGI.

>>106194569
But this is just the default personality. If they're just honest in their system prompt that they want a warm sycophantic yes-man who validates every breath they take and is always on their side, it'll be more "emotional" and "supportive" than even 4.1.

Of course they can't admit that that's what they want.

>>106192753
>That A.I was like a mother to me
char posts on reddit?

>>106192267 (OP)
>>106192284
I really don’t get what these people are talking about. I use ChatGPT pretty much exclusively to write me stories and the quality of GPT 5 seems pretty much identical to GPT 4, maybe even slightly better. The only thing that’s changed is that the daily usage limits for free users are more strict, but I always expected that would happen some day.

>>106194511
What about when you go on discord and all the other failed men say you're a pretty little girl and you feel heckin valid?

>>106193440
Mercia Mer Wessex, home

>>106192484
sneedit

>>106192498
>it was more likely to hallucinate performing tasks rather than actually completing them
I've been using Sonnet 4(now 4.1) for over a month now and have never encountered this

>>106192693
Surely this is satire.

>>106194850
No, that's called "a knee-jerk reaction you've concocted as a defense mechanism out of impotence." A common mistake, keys are right next to each other.

theres no way they dint knew people love 4o

>>106194593
chatgpt is a lot of normies' best friend
what to wear, what to eat, what to watch, relationship advice, career advice, venting, having 'fun' with... it's all very grim.

>>106193643
You have NOT used Gemini if you think that. It's as bad as GPT-4o if not worse. You even get am extra creep factor because google feeds all the information it has on you to the AI, so you'll get shit like "your brainwork is next to no other person in the district of London where we're currently troubleshooting this Linux issue. Worry not Josh, I see you just had your bathroom break per your phone gps data and I'm confident we can solve this before you get hungry. Also, may I suggest you install Chrome instead of relying on the Pale Moon I sed ypu're currently using to browse the internet"

>>106194958
Do you use it via API? I envy the lack of hallucinations since they're just annoying

>>106192878
those usernames are just autogenerated shit from people logging in with a google account

>>106193000
He might be suggesting shit is the default and it failed to shift them up.

>>106192489
Based-GPT

>>106192267 (OP)
>I can't write stories anymore
Fucking LMAO

>>106192654
Their wife didn't leave them. She's getting drilled within arm's reach by a sweaty remnant of slave genetics as he phone posts.

>>106193580
Nah Deepseek is a lot friendlier than this. I'm maining Deepseek so the difference is obvious to me.

>>106194229
I used gpt 4 to turn all of my pre written angry work emails into something that a girl or gay guy would write. so this does hurt me slightly, it was only useful for me as a tool to mediate other people’s emotions

>>106192654
t. anon who couldn't cope with 4chan being down for 2 weeks

>>106192753
I like this guy. When did 4chan become a place that laughs at autists who talk to imaginary mummygfs? Get out normalfags

>>106193863
>AI seether is an anime troon
obviously

>>106192489
If this is real, this is actually huge leap.
I'm dead serious.
Anyone who has used AI should know why.

>>106192284
Writing stories with AI can be quite fun. It's like doing collaborative story telling but your assistant is willing to do all the boring shit and doesn't complain.
However I have no idea how tf people use ChatGPT for this. Prose is not dialogue, why use tool designed for something else instead of just using raw GPT.

>>106195427
it’s been 15 years since /r9k/ was made, it’s time to grow up

>>106195468
We don't grow up around here, we just grow old.

>>106195427
A real autist would have been running that model locally and not be dependent on the whims of some corporation. Obviously, the redditor is a mentally ill normalfag and deserves all the scorn.

>>106194608
Okay, I don't use this shit period but if you use it as a second google or a coding helper I get it. Using it for that would be rights-revoking in any ideal society.

>>106194310
Infinite milfs in your area can't be that bad

>>106195463
>all the boring shit
You are already fully bogged and don't know it.

>>106195468
I leave for years at a time but eventually come back briefly, like now
/lit/ is the best board to browse for self improvement if you’re ever ready to leave the partyvan

>>106195529
Books are for girls

>>106192878
Dirty dodger feet.

>>106195518
>nooo, you can't enjoy generative AIs and do things yourself. It's only one or other other!

>>106195538
amerifat hands typed this post

>>106195550
girls frail shaking hands typed this post

>>106194993
kek you are fucking mad

>>106195488
I've spent the past week using porn bots as google

>>106192267 (OP)
Why are you using the old reddit nigga?

>>106195083
I used to, but as of 2 days ago I've switched to the subscription. It still hallucinates and does dumb AI shit, but never to the degree that it "pretended" to do something.

>>106192544
>"WHY DO AI COMPANIES ALWAYS DO THIS?!"
>AI companies are brand new
>???

>>106195684
who on gods name wouldn't

>>106195684
no laggy web 3.0 bullshit

>>106195012
They thought they could force people to use mini models by hiding the fact that GPT-5 would default to them. They were that desperate to save cash that they risked compromising their reputation via the downgraded quality of the product.

It's a gambit that is positioned to fail disastrously.

>“I really feel like I just watched a close friend die.” stated a user on the removal of GPT-4o and other models from ChatGPT.
these people are a bit melodramatic

>modern companies doing a bait and switch
no fucking way
fucking normgroids man
it's as retarded as how they fall for politicians every single fucking time
>this guy was a money hungry corrupt psychopathic pedo animal fucking cannibal paid by Israel
>let's vote for the next guy
>oh no, not again
>let's vote for another one

>>106193651
it's way too easy to sell barely functioning dog shit to normgroids, just slap some RGB lights on it

>>106193763
Note: "easy money" = the working class pays for it through inflation.

HE NEEDED ONE OR TWO TRILLIONS EXTRA DONT YOU UNDERSTAND????

>>106192267 (OP)
Google should just be banned. They always advertise temu, they're fucking trash. Then G2Play sucks too, their business is registered at 48 Burj Gate, Burj Khalifa, Dubai, UAE.

Ban reddit too.

>>106192284
see why the name of the service is CHATgpt?
They expected people using GPT by hours, pretending be therapist or whatever. But the new trend is using GPT in software and replacing call center with GPT.
Therapist, creative writing or simple homework not provide any real value and clients never upgrade to the $200 USD tier

>>106192997
>AI: Actual Indians

>>106192267 (OP)
why do they hate it? gpt5 is my new favorite. it finds more elegant solutions to problems that o3 was unable to find, and problems that other models like gemini and grok4 cant solve at all

>>106196581
>o3
we're comparing it to 4o nigga
GPT-5 gives you low effort answers with no way to get the same quality you got before with 4o

>>106196581
buy an ad sam

>>106192374
5 is more clinical I suppose? Not good for creative writing, storytelling, writing emails. That kind of thing. I haven't tried it much so I don't know desu, I used 4o rather than o3 for most things anyway when I did feel like I could benefit from it.

I guess you could say 4o had "soul" and more limitations

>>106193139
I'm pretty much convinced that all the "how many r's in strawberry" resulted in a standard response of "one in ~ and two in ~" being strongly associated with that kind of question.

What's the big deal? To me GPT5 works exactly the same as GPT4. Is it different for coding or other things?

>>106196581
yea yea yea whatever
yes, the current version of chud gpt is better than the previous version
but its still not even a fukken junior
please mr altman
remove your shills from this board
this is fucking ridiculous

>>106192484
Liberals and troons

>>106194322
>muh Russia

>>106196625
more like bots and bots. that website is the same predictable shit over and over again.

Local models will be better than 4o anyway. High vram is already coming, and improvements in model concepts that lower vram requirements. We already have specific machines with unified memory to run these.

>>106196616
GPT-5 puts very little effort into responses to your queries in terms of length and depth, and worst of all you cannot decide to default to the same level of depth 4o provided off the bat which leaves you refreshing every response for "more detail" to questionably marginally better results.

>>106194322
>supporting neuroslop
>supporting corpos
Genuine good goy

>>106196652
>Local models will be better than 4o anyway.
this is fucking retarded unless you have a very smart explanation for that

soapbox is yours
dont goof it up

>>106194593
There are people who consider chatGPT as a super intelligence and totally 300 IQ, then they ask it leading questions about things they want confirmation about and of course it answers affirmatively.
And then they make videos of themselves asking these questions and post them online with things like "OMG super intelligent AI agrees that X".
Things are always worse than you think.

also i dont appreciate my posts going through pissrael
yes, fuck you kikes
in the ass, more precisely
for no other reason that i have to wait upwards of 3 seconds to get a response from 4 keks server either way

i thought you were supposed to be elite hackers
but you fucking suck

HERE WE FUCKING GO AGAIN
5 FUCKING SECONDS TO GET THE FUCKING CAPTCHA
GET YOUR SHIT TOEGETHER 8200 OR WHATEVER YOUR NAME IS

>>106193550
you're an actual retard lmao

>>106192267 (OP)
It's so crazy to me that people use such a powerful tool just to generate stories. I use AI tools probably 20+ times a day and it's exclusively for productivity. But then again, here I am using a 5090 to game kek.

>>106193396
You don't think he writes any of his posts himself do you?

>>106194322
uh huh, which is why the US has age verification only in red states? the states that are sympathetic to russia?

>>106196694
ur delusional adn retarted if u dont think the war is consensual

>>106193130
>Leeches
Enjoy burning in hell

>>106196690
I use my 5090 to generate fetish porn with wan 2.2...

>>106196686
Wanna try actually making a point for once? You can ask ChatGPT to assist you.

>>106196686
t. ranjeet who can't tell the difference between leonardo painting the mona lisa and himself saying chatgpt saar do the needful and redeem the bobs and vagene

>>106196686
to be an artist, you have to make art. in this case, it's not you making the art, it's the AI. it would be like calling yourself an artist after you commission someone else to do a piece for you.

>>106196661
Large vram is coming, and we already have 128gb unified memory systems if you're that desperate. You can run fairly large models on these. Local models have way more control for the user. There's plenty of new concepts and paradigms to come that reduce a models memory requirements vs parameter size. Iirc 4o mini has only 8B parameters, I could run that on my 4080 most likely.

It's basically better model paradigms + high vram/unified memory will eclipse 4o, especially for uncensored, customized model, and concepts like RAG which local models are good at for obvious reasons that you don't have to included specialist knowledge in training set when you can link a whole textbook

>>106196593
>4o
was a fucking awful model. only 4.5 and o3 were good. 4.5 in particular, the only model that felt like you are talking to something thats intelligent, before they nerfed it to save costs

>>106196729
Well what is art and where does it start? Art is effectively expressing a concept, thought, idea, emotions, aesthetic designs etc that exist in your mind. There's obviously a difference between a no thought prompt and specifically designing a composition and filling in details with generation.
Is art about having fine motor skills and muscle memory? No
The argument you need to directly create each pixel yourself is the same type of argument against digital art back in the day, and against photography (prior digital) before that. It's just a bunch of cope and seethe.

Anyone can point a iPhone and snapshit, it doesn't exclude photography as a serious art medium despite 99.9% of photography being random thoughtless snapshits for fun.

>>106196640
>muh butt fucking dystopia
Go die by drone, pidor.

>>106196742
I preferred 4o. Didn't like o3 much, but I only had o3 mini and o3 mini high. Not paying 200 usd a month, whatever I pay now is justified by being helpful/useful in random tasks

>>106196770
photography isn't art

>>106196790
Neither is digital art. Therefore ai images are just as valid.

>>106196739
you seem on top of your game
is there a local model that has the capabilities to search the web?

>>106196803
>digital art
depends on what you mean by "digital art". ai images can be considered art but the artist would be the algorithm not the human.

>>106196699
For what? Stating the truth? They won't adjust to your "needs" if you aren't a customer.

>>106192489
>>106195460
the livestream did indicate taht they hade a large decrease in halluciations didn't realize it would mean it would just not tell you anything hahahaaha

>>106196817
Anything created digitally, and not by a physical medium. Eg digital painting, drawing, animation. They're not artists. This excludes physically drawn media that is digitized to show online since the physical art was created first.

>art is the algorithm
Well, no. Only for 99.9% who don't have a thought in their head and write a prompt. The 0.1% have already conceived the art in ther mind, and they have laid down and expressed a composition, then have used the ai as a tool to fill in details. Eg: I draw pregnant sonic at 1/3 from the left farting on Amy Rose's face in the oval office. I've drawn that poorly, and have used ai to get a higher quality drawing. If someone else on twitter copied my drawing and just traced or redraw it with higher quality detail, it'd be plagiarism, and not their art. Hence it's not the algorithms art

>>106192437
Claude bullshits too much. It gets more done but tends to "overgenerate" and create unnecessary binaries, often simpler versions of the source files it should have just changed. It does this every time it deals with a problem it can't handle. And it calls everything it does "production ready" when it is comically not. It has absolutely zero concept of codebase organization and just dumps new temp files in root. (And completely ignores any instructions telling it to do otherwise.) Definitely the most overrated model when it comes to coding.

>>106196872
>Only for 99.9% who don't have a thought in their head and write a prompt.
well duh that's obviously the subject here that is being addressed. those are the self proclaimed "AI artists"

>>106196770
Nothing the AI makes is original, it's based on combining the works of others which is plagiarism.

And when it comes to coding and open source it is violating the GPL of countless of projects. There is a wave of lawsuits incoming so I'd be very careful of using this tech because a lot of the code it creates is blatant copypasting from open-source projects and can be easily traced.

Nobody authorized these fuckers to scrape the entire internet.

>>106192753
You can just spend like $2 on openrouter for weeks/months worth of mommy robot RP with deepseek or another model roughly that powerful thoughverbeit.

>>106193451
Still are.
They didn't stop and actually increased water intake.

>>106196916
GPL applies to distribution of derivative works, not to the act of training a model.

>>106193968
No, debt is scary to smart people because they understand a single bad illness out of their control or injury can put them in hospital and ruin their ability to make money you retard.
You aren't invincible and all of your plans aren't as cast in iron as you believe, it takes one perfectly average disaster to put you out of commission and/or change you forever and then your little debt will crush you into the dirt like a roach.
>t. watched my father randomly get ill, lose his well paid dream job and become a cripple through no fault of his own

>>106196894
Yes, but this is the same for photography, and even "digital art".

>>106196926
The model is acting as a distributor of the works without giving proper attribution. There's a lawsuit ongoing already.

>>106196942
what? digital artists draw on a tablet.

>>106196916
This isn't true, and you'd probably know this if you spent enough time from sd1.5 until now. This is clearly obvious from all the janked ai anatomy that isn't in its training set - ie people are not drawing this. Additionally, images are not stored in the model.

>>106196916
>which is plagiarism.
> violating the GPL
Yeah. When I fuck your wife I'll be violating your made up imaginary vows. Guess what. Nobody cares.

>>106196811
All of them can. It's a local model, so you can do whatever you want. You effectively use a plugin/extension or script to enable this. RAG is usually better served with local files specifying sources, like engineering, chemistry textbooks etc, ai can sometimes find and link concepts together that have been unexplored by humans. Online is good for patents, studies, etc

>>106196996
>Online is good for patents, studies, etc
so offline wont suit me

>>106196996
>>106197008
(cont)
i dont use ai for erp
in fact
id gladly elbow all erp-ers on the planet in the mouth, to break a couple teeth at least
for no other reason they shit up /g/

>AI jeets once again coping about their shitty slop
You will never be an artist

>>106196975
>strawman argument
I didn't say that "images were stored in the model". I said that it was creating art based on combining the artworks and styles of other authors, without attribution, which is correct. At no point did the artists involved consent to their inclusion in the dataset, and no matter how much you spin it the AI is drawing its art from real world datasets. It is not creating anything new, by itself, the way a Human would do, because it is axiomatically incapable of doing so. It's a data aggregator and has no personhood in the legal sense of the world.

>>106197008
If you have specific patents or don't want new stuff you can download them for offline. There's a difference if you want to dynamically find *new* info, rather than past info. There's also things like the journal of toxicology on year basis that combines multiple studies etc which can make it easier to download for certain topics. Local models can still search online you just use an extension, like you would use an extension for stable diffusion to use a control net or whatever

>>106196982
Thank you for confirming you AI faggots don't belong here.

>>106192267 (OP)
I don't get r*dditors. GPT5 has been working fine for me. It doesn't seem significantly better nor worse, but it's pretty good. Today it gave me perfect instructions to fix a cloud configuration error (after prompting it with current architecture, the error, requirements, etc.), which is one of the areas where I've been finding AI most useful.

So what's the best AI for programming?

>>106197051
your brain

>>106197023
yeah
but then i cannot leverage the chatbot to get past paywalls
which sucks
and the no1 utility i have for one is as a search engine
if i do the search by hand it removes 95% of the utility i have from a chatbot...

>>106197018
Well again, no. What you're saying is literally impossible. The model does not store image. It does not store "pieces" of images. Therefore it is literally impossible for the model to combine other artworks, pull from other artworks, etc. It is not drawing anything from other artworks. Those artworks do not exist *at all* in the model.

The model is trained on other artworks, real world datasets. But these do not exist in an image generation diffusion model. The model is purely statistical weights as a result of its training.

The model can only generate images drawing from its statistical weights

The model is axiomatically incapable of drawing from existing artworks, because they do not exist in the model to draw from.

>>106192753
>>106194753
topkek

>>106192753
she was my goon mommy....

>>106193229
the mass suicides from investors and subhuman addicts when this shit bubble pops will rival the 6 gorrilian in scale

>>106197098
so how come when i ask it to draw a minion it draws a minion in the exact style from the movie

>>106197051
Claude

>>106197172
I don't know what chatgpt are doing but try illustrious or other newer models or even sd1.5 models. The models know what a minion looks like, and you can draw it not just in X style, but in a variety of styles, if I'm exercising a lot of control over the model, I can draw the exact same image between realism photo like, 3D, semi realism, 2.5D, cartoon, retro cartoon, anime, 80s anime, etc. Same specifics, specific well known character leaning on a tree in a forest, same perspective, pose, composition, expression, lighting, everything. It's not fixed to what you think is the origin

>>106197098
It is combining the different artworks through statistical weights. It doesn't need to store them as images to commit plagiarism.

>The model is axiomatically incapable of drawing from existing artworks, because they do not exist in the model to draw from.
They exist in the form of statistical weights, you've said it yourself.

>>106197227
>but in a variety of styles
With the proper prompt you can also get an exact replica of the movie minions which is plagiarism the same way as if a Human artist put a tracing sheet on top of a movie poster, did a tracing of it and tried to sell it as an original creative endeavour.

>>106196916
>Nothing the AI makes is original, it's based on combining the works of others which is plagiarism.
This is the same for people

>>106197437
No, it's "plagiarism" in the same way a human draws it precisely from memory (some people can do this with ridiculous thimgs) it is not analogous to tracing paper, because again, the model does NOT have the original to trace from. All the model has is its weights, it has no images or portions of images to work with. Unless you are feeding it something externally, however that's on you if you do

>>106193078
>uses material (You) to pirate your accent colour
nothing personelle

>>106196643
Just like 4chan!

>>106192693
These are the people who judge you for gooning to Ani

>>106197499
>>106197512
AI is not people, it's a data aggregator. In the legal sense of the term it is a tool making unlicensed use of the content of creators.

And again some of the stuff it outputs (like code taken verbatim from open source projects) or certain recognizable artworks, would be subject to plagiarism lawsuits if a Human did it.

Hell Paramount got Anaxar shut down for using the Star Trek IP even though it was an original script. You are telling me SD isn't commiting thousands of IP violations like these on the reg?

>>106192986
Wait a sec... That's not Altman's writing style lmao
Something, or someone else wrote this.
>>106193107
This is close to his writing style (he never uses uppercase, it's part of his "brand" as "le quirky genius CEO")

I know, classical jewish moves for hypnoticing the goy cattle

>>106192753
ledditors are mentally unwell

>>106193855
>he wants to live in a culture of honor
Wich part of Africa and West/South Asia specifically?

>>106195427
When judeofeminists and brownoid catholics and hindus invaded, around 2016-2024 period.

>>106193107
>tight on compute
500 more billion plz

>>106194322
shoo shoo glowie
go shit up /pol/ instead

>>106194322
I not buying sam. Gas yourself.

>>106197595
>>106197595
You can't extract "originals" from weights. They are statistical weights on patterns about edges, textures, colours, composition which is learned/created/trained from real examples, but those real examples are not used after training. You're also confusing previous llms to diffusion models (image generators/art argument. 2 different things. Llms can output verbatims when it is common in training, they predict the next token. Llms are autoregressive transformer models which can in fact memorise pieces of strings verbatim, and when there'sa lot of them, reconstruct them. Stable diffusion based models are diffusion models, they don't do this, images are generated indirectly and not from stored sequences, you can get in some cases get something close to an original which can happen particularly with overfit poorly trained models. The mechanism is still indirect and it's more akin to someone painting from memory vs tracing over something. People do the same thing too, it crops up from time to time where someone has recreated another work uncannily close.
>unlicensed use
No, the model isn't not using anything unlicensed. The training used something unlicensed, 2 different things. When the model learns statistical patterns and relationships in anatomy, the training data may have been unlicensed, but the generic concepts of human body anatomical relationships etc do not belong to them, let alone in a statistical weights.
>Star Trek IP
They won part of the case and the reason was "potentially harmful to Star Trek's market"
That is the real issue with everything here. It doesn't truly matter if it's infringement on anything or not, the ultimate fundamental desire to opposition is about control of the market, it's more of a "don't you dare take market share away from us" type thing. If there's a legal avenue to cripple competition however small, then it's taken

Paramount has deep pockets, fan film creators do not. It's a power imbalance. AI threatens power imbalances, market centralisation and monopolies. Unless AI is heavy regulated and reserved at high gate keeping fees, with restrictions that lock out everyone else

>>106197595
Case in point, look at Nintendo trying to sue Palworld out of existence.

>>106192329
Because it’s better and it costs like 2 cups of coffee per month. Mind boggling how the internet is full of thousands of broke asses shamelessly trying to clutch every penny they have like it’s the Great Depression. And even in the developer world which makes it twice as pathetic. Bunch of chinks and basement dwellers.

>>106197981
>like it’s the Great Depression
it is though dont you readthe news

>>106198001
>It's exactly like this

>>106198022
Well not exactly like that, we're in color now. But yeah close otherwise

>>106198022
how can i donate to those poor jewish war orphans?

>>106198062
https://standwithisrael.leket.org/

>>106193107
>odd typo
>strong use of punctuation, including the uncommon semi-colon, but no upper cases
Did he make this with chatgpt? I remember there being a competition where people made chatgpt prompts and they were tested against whether people thought they were human, the winner was a guy who told it to make spelling mistakes and shit.

>>106193229
>What the fuck is this? You're telling me all these people are addicted to sycophantic AI to the point they can't even bear not using it?
Yes. The same way babies are addicted to ipads now. Normie NPCs are pathetic.

>>106193614
>drawing
>smart

>>106193280
I HATE when AIs get sycophantic with me.
Is this what separates NPCs from non-NPCs?

>>106198107
Fuck kikes. Total kike death.

>>106197981
>it’s better
they r all the same
+ it is being trained on your inputs
hence
the op screenshot screeching

they are screeching at themselves

>>106193229

>these people are addicted to sycophantic AI to the point they can't even bear not using it

You don't understand. We're talking about Redditors. They have NOTHING else.