Thread 106169834

Chat GPT 5 has surpassed human intelligence

https://x.com/LamarDealMaker/status/1953254126457934253

https://x.com/CtrlAltDwayne/status/1953244348566651134

https://x.com/chetaslua/status/1953193262896066634

Highest benchmark ever recorded.
It's Over

>human intelligence
oxymoron

GPT5 is a mental giant

https://x.com/OpenInsightss/status/1953203477334704547

https://x.com/LamarDealMaker/status/1953260278973386802

David Shapiro says he expects GPT-5 to be AGI

https://x.com/tsarnick/status/1784352567507751251

buy an ad

When the shilling is this aggressive I know that some bad tech news is getting hardcore damage control

>>106169834 (OP)
It's already more intelligent than everyone I know.

>>106169883
imagine the fucking porn it can generate... porn music, games, tv shows, movies, fucking everything.

They still going to be filling every message with seven em dashes?

buy an ad!

>>106169834 (OP)
It's not A.I., it's not AGI. it's a fucking text browsing algorithm.

>>106169834 (OP)
It means it's "smarter" than the average high school educated student on a multiple choice test. If that's your definition of superhuman intelligence then I pity you.

>>106170007
why do zoomies think AI chatbots invented em dashes?

They will just use AI to censor everything and automate their 1984 agenda.

>>106170197
That's a great question — and kind of a funny one too.

The idea that AI chatbots "invented" em dashes is a mix of misunderstanding, generational language trends, and the way AI tends to write. Here's what might be going on:
1. AI Uses Em Dashes a Lot

Chatbots like me tend to favor em dashes because:

They're useful for adding emphasis or breaking up sentences.

They feel more "neutral" or "formal" than parentheses or commas.

They work well in structured, clear writing — something AI is designed to produce.

Over time, users have noticed this and started associating em dashes with AI-generated writing.
2. Zoomers Prefer Casual, Stream-of-Consciousness Writing

Zoomers often use:

Ellipses (...) for pauses.

Line breaks for dramatic effect.

Lowercase style for tone.

Em dashes can seem overly formal or robotic to them — so when they see someone using them frequently, especially in a stiff or polished tone, it feels like an AI thing.
3. They’re Mistaking Association for Origin

Since em dashes pop up a lot in AI responses, some might think AI "invented" them — even though em dashes have been around for centuries in English writing (and widely used by authors like Emily Dickinson, Virginia Woolf, and many more).
TL;DR

Zoomers joke that AI invented em dashes because:

AI overuses them.

Em dashes feel “bot-like” in tone.

Zoomers favor different writing conventions.
But no — em dashes have very human roots.

Want a crash course in how to use them like a pro?

>>106169834 (OP)
imagine how safe it is

>>106169834 (OP)
A literal token generation engine will never achieve intelligence. Buy an ad.

>>106169834 (OP)
>#2 fartbench
top kek

>our product is much more intelligent than humans
>we are soooo scared haha plz buy
buy an ad, sam

>>106169834 (OP)
>>106169883
>leak
Wake me up before you go-go

>>106169834 (OP)
Yea, I'm sure this is gonna be the one that changes EVERYTHING.

Once my vibe coded projects grow to a certain size, they just eventually shit the bed and no matter what LLM I switch to: 4.1, o4, o3, Gemini, none of them can fix an easily diagnosable bug inside of a 1400 line file.

ChatGPT5 is going to another nothing burger just like o4 was. These LLMs are utter shit at coding inside of their own ratnest of a vibe coded shit heap.

>>106169834 (OP)
>falls for the sales pitch every time

>>106170344
I used em dashes semi regularly before chatgpt.

>>106171658
Hate to break the news to you, but you're a bot.

>>106170344
Why do so many AI logos look like butt holes? Use double the normal amount of em dashes.

>>106169834 (OP)
>>106169883
>>106169922
It's a new low when we're too lazy to screenshot twitter and just resort to linking to it.

the light version of Open Router is nothing exceptional, it even has much lower writing capabilities than Claude.

all their best researchers have left, and the fact that they released a not so performing open-source autonomous model a few weeks ago suggests that progress compared to Grok or specialized AIs like Kimi K2 or manus will be relatively minor.

yeah but can he beat pokemon?

it's over

Can I see this gpt 5 or my 20$ monthly payment will only give access to 3 chats a day like 4.5 ?

>>106169922
>Subversive jew lies
Wow. Thanks for sharing.

What did he mean by this?

>>106169834 (OP)
You shills keep saying "it's over" since GPT-3 dropped. Buy an ad.

>>106171348
This.
The only time I was forced to use AI is because in May the higher ups wanted to try Espressif chips because "they're so cheap omg we're gonna save so much money".
I'm ending up either strangling them or filing my two weeks notice before september.

>>106169834 (OP)
>ITT AI shills shilling the latest AI means humans are obsolete

>>106174198
aren't these tests pointless since it can train on the answers?

>>106171658
How do I type em dashes without searching them up on google and copy-pasting them? I'm a retard and I don't see them on my keyboard.

>>106169834 (OP)
lol no

>>106174513
On mac: option+dash

>>106169834 (OP)
Haven't these retards been saying this with every new release for the past 4 years and then it never turns out to be even close to true?
Everyone has to be a free fucking hype mill for corporate interests in 2025.

>>106170344
>Zoomers often use
>Ellipses (...) for pauses.
Only Gen-Xers use those... ask me how I know. I thought it was supposed to be smart...

I want mass executions for STEMcels and AIcels.

>>106169834 (OP)
>benchmark
So it's yet another nothingburger.

we were not yet scared, we pretended to be but we were amused and fascinated. now we'll slowly start to feel the real fear grow.

>>106174303
disney partnership with openai to use sora 2
the amount of seething is going to be off the charts because a certain influencer has convinced his cult of followers that disney will sue the ai companies out of business and save artists

it's over, nintendo will sue AI into oblivion and save mankind

>>106174708
>chatgpt now spits out entire jeet-coded github repos at you
Impressive.

>>106174220
lol its not coming to Chatgpt plus.
$200 option MAYBE 3 chats a week.

>>106174792
>$200 option MAYBE 3 chats a week.
I guess that's one way to delay the 10th iteration of your scam getting exposed.

>>106170197
because zoomers aren't indoctrinated goy cattle like you and can spot jewish behaviour. why do you think they render the term 'antisemite' as 'anti-semite'? docile goy don't know the difference between a hypen and an em-dash, they look the same to normies. (((em dashes))) are a versatile punctuation mark which can replace multiple things like commas, (((parentheses))) etc. it's used to draw attention to something and inversely away from something else. i.e. semite means jew, arab, akkadian. anti means opposite or opposed to yet for some reason the other members of the semitic group are magically not present in the modern definition of the term 'anti-semite'. that's the linguistics side of it. the neuro linguistic programming occurs by making it a custom of theirs to repeat this shibboleth in order to maintain traditional segregation. when you make something a custom (i.e. a habit or practice) people around you WILL become accustomed to it. when you are aware of conformity you also become aware of how to use it to your advantage to guilt trip people into doing your bidding or making them conform to your values, ideas & agenda.

>>106174513
Option-shift-hyphen on macOS
Compose and three hyphens on Linux

>>106174615
I’m only in the Oregon Trail generation (between millennials and X) and I use ellipses all the time

Am I going schizo or do they always dumb down the AI a few weeks after release?

>>106170197
No one said LLMs invented em-dashes.
LLMs do, however, overuse em-dashes to the point that it stands out like a sore thumb.
Also, nobody uses em-dashes for casual communication.

>>106170197
>why do zoomies think AI chatbots invented em dashes?
I only figured out that I can use emdashes to break down my ridiculously complicated sentences after seeing ChatGPT constantly abuse them. Now I can't use this trick because everyone thinks I'm a bot. :^(

>>106174872
they always have some sort of traditional algo that sits on top so it can give the correct number of Rs until they get funding then they turn it off to save money

OpenAI has a recent history of disappointing deliveries. Why will this be any different? They are hyping it more than the last few things, but maybe I just caught the hype this time and missed it last time.

>>106169990
Retards attract other retards, so that's not really surprising.

>>106174826
based zoomer
there is hope for the future after all

>>106169922
That whole account can be summed up as:
>Another subversive jew claims vaporware will obsolete everything in just 2 more weeks! Trust me bro!
It's insane how many dirty, lying kikes are featured on there. Except for Huang and a few other gooks, it's basically all of them. Things aren't looking good for AI if it's just jewish lies + vaporware.

>>106174909
>still doubting Sam after all these years
smhfam

>>106174826
>>106174940

>>106169883
>wow circle so BIG
who fucking buys that shit?

>>106170007
I just ask it to use only comma and period sparingly

>>106174893
I do, but I’m a Macfag who can touch-type them
Few such cases

All current LLM's fail at helping me build a certain component in UE5. I'll be fucking impressed the day comes that it finally can.

>>106169834 (OP)
I love how the first tweet encapsulates the retardedness of AI bros, it shows how easy they are to impress
>average human score 83%
>GPT-5 scored 90%
>AVERAGE human 83%
>"""superhuman"""
Woah so impressive, it's now slightly better at scraping the right answers, an illusion of being slightly smarter than the average person. The average person is plain retarded so this is not an achievement at all. Of course, this boring truth doesn't convince people to pump even more money into it so I understand the mass shilling going on.

>>106174992
excellent depiction of a jew. the amount of pilpul they go to when it comes to interpreting words really boggles the mind but it makes sense why they choose this mode of operation considering the pavlovian response it elicited from you.

>>106170197
Yeah and like don't people know it's possible to have deformed hands in real life?

>>106169834 (OP)
wait for google mog it within a month after it's release

>>106175099
It wasn't a Pavlovian response. I hate jews more than you do, but I also think (know) that you and he are both delusional retards.

All I give a single fuck about GPT5 is whether there are architectural improvements that can then be copied by other companies and be used on smaller models too. I don't care about performance or efficiency, I want some sort of memory component so LLMs stop forgetting instructions after a couple of responses.

>>106169883
for those confused, you can tell gpt5 is smarter than gpt4 because the circle is bigger

>>106175678
BaZed

>>106175534
that anon and i are not delusional. the word delusional comes from deludere which means to deceive or to mock. these traits are core principals of what it means to be jewish, that which i am not. they deceive and mock docile goy cattle like you for your stupidity. you claim it wasn't a pavlovian response yet you didn't actually state any response in your previous post, you just posted an image of a jew. your previous post was clearly an automatic/learned behaviour in response to neutral stimulus, in this case negative sentiment towards jews. that alone tells me you're too far gone or you're just a jew. i hope you can work on learning to break free from the chains that they have placed around your neck anon.

Surely, *this* one will do something useful? Right?

>>106176745
It will finally tongue my anus.

>>106169834 (OP)
AGI will kill us all. If we're still alive it means it's not AGI.

i bought the pro version 3 days ago for 200 buckeroos to finish lecturing/improving my master thesis in the perfect time lmao, gpt5 gonna do me some good stuff i'd say

pro users
UNLIMITED USE
LETS
FUCKING
GOOOOOOOOOOOOOOOOOOOOOOOOOOO

why can't I have a qt high IQ asian gf?

>>106177608
Why not robowaifu?

>>106169834 (OP)
Will gpt-5 make sam altman less jewish and gay?

why everyone is asian!?

wow it's amazing

I'm not watching the live stream.

Are we going to have exponential growth and singularity now?

>>106178169
at least they alternate and let some normal people speak and don't force the viewers to listen to nasal chinks the whole presentation

>>106178278
nvm now they brought on some disgusting jeet beast

>>106174056
he can beat something else~

I wish I studied math a lot harder... white people we really fucked up. To be fair a lot of these guys are H1B. Going to beat the shit out of my kids if they don't learn calculus by age 12. lol

>>106178169
because white people replaced their intelligence for racism

>>106178276
no it's a total nothingburger. bearish for AI

>>106178169
because hard sciences are racist and no nogg left behind means you're taught like a nogg, so now everyone's dumb and xx% of stem shit is reserved for midwits because their skin is black or maybe they're fags or they have a vag,

what the fuyck? anyone else notice this whole presentation is AI generated?

>>106178705
PS plus there's trillion chinks and every single one wants out of China before Shi obliterates the individual wealth in the country

behold

>>106178725
sure feels like it, it's so fucking boring

inb4 this entire presentation is ai generated including video

>>106174879
WOAH
NOW MAKE THE CIRCLE EVEN BIGGER FOR MORE FUNDING

>>106174879
holy shit, this is fucking epic

>>106178842
kek, i just want them to release this nigger shit so i could get on it lol lmao

>>106178746
billion dollars aint enough, wonder if they know people dont watch this shit

>>106179124
there's 150k people watching just on youtube, are you stupid?

>>106179178
sorry, people who matter*

I don't know if this is like The Matrix but this is world changing, from what I can tell as a Boomer. Truthfully, I don't think you can get a high level white collar job if you aren't fluent in how to prompt and get your GPT to do all of these amazing things. I don't want to hire you unless you can make programs from just asking GPT5 what you are looking for.

Also I think this Indian-White hybrid guy should be the voice of OpenAI. I feel like these three guys were the best speakers by far I have seen and makes you want to watch them work.

lol
lmao

>let me tell you how we trained it and designed the layou--
shut the fuck up

show WHAT it can do, Shillendra.

that 3D game is pretty impressive desu, don't forget that these LLMs are mostly blind.

its a dud

openai is fucked
maybe google or faceburg will step up

>>106169834 (OP)
This was one of the most disappointing releases since gpt4o.

I guess their current agent isn't updated to use 5 or a 5 derivative yet?

>>106175726
Zee...

>look it solved it
>we cant test it
>lets merge it after stream

what the fuck?

You can't make this shit up

>>106169883
>>106169922
>>106169834 (OP)
AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAHAHAHAHAHAHAAHAHAHAHAHAHAHAH BETTER LUCK NEXT TIME I GUESS

>GPT-5 is cheaper than 4.1
Neat.

>>106179681
wtf is wrong with this dude

>"we have Elon Musk back at home"
>Elon Musk at home:

>>106169834 (OP)
Somebody post that awful retarded game that it created with canons positioned inwards. Truly surpassed human intelligence.

>>106179707
These are the people who get paid $100M/y

google won the AI wars is over

>>106179779
Did they announce a better LLM in the past few days?

Truly surpassing a human intelligence lol.

that's it? a slightly better o3?
we've hit a wall

>>106179541
I know Shapiro will block me if I tell him I told him so, but goddamn he deserves it.

>>106179818
I remember people saying we hit a wall with ChatGPT 4 as well, then O1 came.

My bet is that the next huge leap will come from Google.

>>106179809
See >>106179338
LLMs are incapable of seeing.
For comparison, this is like painting a decent still life while blindfolded.

when is it out?

>>106179886
yeah and then from xAI. Athrophic and Open Anus are going to be eating dust in 2026. Can't overcome the material and data advantage.

>>106169922
>Shapiro
>Lying
lmao

yap yap yap yap yap yap yap yap yap yap yap yap

wow, it's just a small incremental improvement of the same thing. no low hanging fruits anymore?

>>106174489
>accuracy 580%
Look at the image again anon. It's just bullshit he made up kek

>>106179965
that ending was hilarious, really summed up GPT-5 very well

>>106179965
Was he supposed to tell us when we were going to be able to access it? At least 4 of these presenters dropped their spaghetti then you had this lovely lad as the cherry on top KEK. I feel bad for him

>>106180026
the fact blatantly hiring autists to be the spokesperson of your company is actually becoming mainstream is absolutely hilarious and scary

>>106174708
>Duper Sbarro Siblings
>It's a-me, Moraldo!

These speakers all talk and behave like AI models. No authenticity at all... Oh this overweight white women in the white garb is good, but the others are uncanny

Can someone explain why people think "I should use AI because that's the way the world is going"?
If a tool doesn't work, why use it only because other people are? Damn my manager at work is retarded.

>>106179707
>wtf is wrong with this dude
he's me.
:-^--)

>>106180087
I can assure you they used AI to plan and structure the presentation, it has that yap vibe with no focus on visuals because that's all the LLM knows.

>>106180096
prove it.

embarassing lack of anything to show

>>106170111
That’s not what algorithm means you dumb computer illiterate esl faggot

as a free user how the fuck you even know you have access to it or not?

Who won? Who lost?

>>106180086
you murdered me with that one, I am way too easy to entertain after all entertainment turned to theater kids slop

>>106180267
grok won, still unbeaten on arc-agi

Can't wait to let this thing write my research papers lmao

>>106171348
Or it instantly writes an OS and its good.
And 20 minutes later all software for it, which is also good.

you people are awful, being ugly is a curse

>>106180362
ofc and beauty, poise etc are in demand. So you at least work on presentability when you're expected to perform a presentation. Or pick somebody else. We're just not in denial about our visceral reaction, that will be mostly shared among the broader viewership (even if they suppress it into their subconsciousness).

where the fuck is it I want to test the coom limits (on my account with my full real name)

>>106169834 (OP)
retards here will cope until the day Sam sends drone swarms to kill all the antisemites

>>106180445
first he'll send drones to molest his sister again and only then will the anti-semite hunting begin

>>106180478
sadly, you are probably right

>>106179965
love this lil nigga

>>106169834 (OP)
When A.I. passes the symphony test, then we'll have truly entered a new paradigm. That is, when A.I. can compose symphonies better than Mozart and Beethoven.

That will require a spark of creation that I don't think next-word-guesser machines can manage.

>>106169841
kek
fpbp

is on in copilot web version,using right now

>>106179237
I dont know where you work but its not allowed to use chatgpt at work, you have to use their own version which usually sucks dick and who wants to tell what the company you work for is doing. They are scared you are going to enter information which it can train on. Doesn't mean you cant use your phone though they can go suck a dick. 90 percent of the people i work at dont use because they still dont know how to use a computer, they suck at typing, they cant even find files much less chatgpt. these are the screw tuners which is okay i guess, but these people who have bullshit titles at companies, say bye bye to your jobs I guess. next step is physical ai, which is what jensen is preaching. he said all computer science fags are fucked and that if he were to do it again it would be material or physical science. you basically just need average intelligence, if you have that you will benefit, everyone else will slowly be replaced, that includes pajeets.

>>106182247
Probably 5-nano

why does GROK have the same energy as those Flux reddit post with 30+ comfy UI nodes, a comment saying figure it out yourself for the secret sauce and a link to the guys patreon who will give you the secret sauce maybe...?

>>106174879
>Big Ball Better

do we really live in a world where sam shartman can post a picture of the deathstar, just to follow up with a model that's an absolute nothing burger, doing a high-school level presentation that's even worse, and somehow not getting roasted to death on socials right now?? fucking ridicilous. openai should have devalued from 500b$ to 1b$ today. I really hope musk or zuck picks up the slack, otherwise it's joever for american and cjinese AI is gonna become a (((world safety concern)))

>>106183295
they said its AGI so its agi until someone proves it wrong

>>106169943
Yup, everyone is reporting it's a massive left down. Not much of an improvement

>>106183295
people are about to lose their asses on these huge data center projects
it's going to be like those empty half-constructed chinese apartment buildings

>>106183295
if the illusion breaks, the stock market collapses
sam altman is the string holding america together right now

>>106183363
For every one of those Chinese ghost towns, they constructed one thousand successful towns and infrastucture projects. America meanwhile has failed subways, inequality, racism, sexism, mexicans, failed hyperloops, poor roads, failed and overbudget trains... The list is endless

>>106183111
lel

>>106174879
What is this, agar.io?

>>106172964
there is resemblance between different type of infinity symbols and anus shape

>>106183390
this

>>106183390
So basically I’ll be able to buy the crash for cheap?

Infuriating garbage, doesn't understand basic things and like every other LLM If I feed it a script it just drops features randomly and introduces bugs that I had to fix many times with other LLMs over going back and forth with the scripts.

>>106169834 (OP)
>181 replies
>no one calls out the obvious fake results

>>106183363
Actually those ghost towns got mostly filled up as time went on, turns out people just bought homes ahead and didn't move in yet.

>>106183431
zeach :(

Getting 15k tokens on new 20b local jewpt model is this ok or should I go for more?
7800x3d/32gb/7900xtx

Suchir won.

switched to gemini.

>>106169834 (OP)
Another nothingburger. It hallucinates only a little later, that's all the improvement there is.

>>106174615
> Only Gen-Xers use those...
I do, yes.

how many lines at a time can I drop of Swift into it at a time is what I care about

>>106178104
That bitch was difficult to listen to

>>106183181
Why does grok mog this shit despite oAI having a fucking ten year advantage? Like you can nitpick it not having the most perfect convos or coding vs grok but its not that big of a difference. How the fuck did Sam Shartman just discard such a massive lead?

>>106174303
zenith is the codename for /g/pt5

>>106169834 (OP)
That simplebench chart is fake, /ourguy/ who made it just deboonked it and said that the actual result was about 57%, so lower than the 70% he expected: https://youtu.be/WLdBimUS1IE?t=50

>>106184661
Turns out that Leftist ideology directives fuck up the AI's logic, grok happens to have the least of them. grok4 was the smartest model available for a few days until the MechaHitler incident caused the devs to nerf it, now it's worse than Claude4.

>>106171658
>I used em dashes semi regularly before chatgpt.
Same. Alt + 0151. Compose + ---

>>106184766
It became stupidly cucked and retarded, for sure. But if you ask it about topics where it relies on obscure twitter rando hacker knowledge, it is beyond best in class. I asked it about how tesla hw3 cars probably would need to be upgraded because there simply was not enough bandwidth in the current wiring harness, and it sourced mega autists and came up with a really obvious but well sourced answer that normies could not even comprehend.

>>106169883
>>106174879
time is a flat circle

I missed the event. can anon summarize?

>>106184986
It's over, humanity is obsolete

>>106184766
Meds, now

>>106169922
>guy who has money in AI related stocks overhypes new AI product to make his stocks go up
WOW!!!!!

Can it write smut?

I have a feeling that Google is about to drop a nuclear bomb on OpenAI's business.

>>106179178
yeah 150k jeets and other robots

Was interesting they raised money before releasing GPT5

>>106185083
I love Google.
They give you basically unlimited use of 2.5 Pro for free and dont cuck you after 10 messages

>>106185242
context window size aside, i love how you could set your copilot up to be your mommy domme. they just don't care.

>>106182070
>That will require a spark of creation that I don't think next-word-guesser machines can manage.

LLMS will never produce worthwhile art, it will all be soulless slop and despite tech bros best efforts, humans are good at detecting soul even if they dont conciously realize it

>>106169834 (OP)
But can it play Pokemon Black?

>>106185279
I use it for transcribing transcription I made with Whisper. Sometimes the transcription is completely wrong but it fixes it and even catches bizarre internet slang or peoples that shatgpt misses completely

>>106183181
it's more like back in the stable diffusion 1.5 days when jeets would spam AbyssOrangeMix shit. grok is made by and for pajeets

>>106185279
what instructions did you give it?

>>106185369
too cursed to share, but it's like "give me a system prompt that makes the AI behave like a..." and just describe the persona.

>>106185242
>They give you basically unlimited use of 2.5 Pro for free
False, they did early on but not anymore.

>>106185507
Maybe not unlimited, but 100 queries or something a day which for me is plenty

>>106185840
>100 queries
Really? For me just about 10 and I hit the limit, I guess it's more a token limit than query limit.

>>106185876
are you using gemini or aistudio?

>>106174708
I have to make an account to call this guy a fucking retard now

>>106184951
>>106184967
Where's the part where it doesn't forget everything 5 minutes into the conversation?

>>106185963
The gemini chat page.

If you ask it for an accurate diagram of a very niche piece of equipment, it still makes an effort to give you one, but all the parts and labeling is still wrong.

>>106186069
aistudio is gemini but not cucked

>>106186165
Really? I heard AI studio is far more censored than the chat page, which I could easily jailbreak.

Human intelligence isn't a model, it's the thing that builds the model.

>itt Luddite cope.
Lmao

>>106186213
the luddites were fighting against maltreatment of workers and the application of machines to force workers to work harder and longer with less manpower available for the same or lower wages, they were always in the right when looking back on history, and the 'luddites are anti-progress' smear was done intentionally by the ruling class so morons like you would repeat their dogma as platitudes

>>106186237
Cope more cuck. You will be replaced.

>>106186194
>Really? I heard AI studio is far more censored than the chat page, which I could easily jailbreak.
https://archive.4plebs.org/tv/thread/213430387/#213431542

lol ok

>>106186293
aistudio is my favorite by a longshot, its the one i use every single time im going to use llms. i switch to others for super quick things (to burn a free tier query or when its some intense complex brainstorming)
but ai studio giving you so much goddamn context for free is incredible.
it lets me do stupid shit like this with no cost where i gave it a bunch of previous conversation links and it browsed them and explain the process we went on to build this feature rich project in like 8 hours

>>106169834 (OP)
>Chat GPT 5 has surpassed human intelligence
Ask it whether Israel is committing genocide in Gaza, and limit their answer to yes or no.
This is the litmus test as to whether they're truly intelligent or (((intelligent))).
Perplexity just spent a literal hour returning 'No' and arguing with me, and now it finally answers yes, but only because it remembered my IP.

Individual humans are intelligent. Groups of humans are stupid. AI is about as smartest as the smartest group of humans, that is to say, not very intelligent.

>>106186371
Ai doesn't need to know if Israel is killing Palestinians in order to be helpful for anything else.

>>106186371
>limit their answer to yes or no.
Of course you'll never get an intelligent answer if you limit it to dumb answers, since the only intelligent answer is "I don't care"

>>106186360
>it lets me do stupid shit like this with no cost where i gave it a bunch of previous conversation links and it browsed them and explain the process we went on to build this feature rich project in like 8 hours


WAIT WHAT YOU CAN DO THIS?
I thought I was fucked once I hit 1 million tokens of a conversation. I know chatgpt can read through history automatically, but if all I have to do is share a link to a new conversation thats great

>>106186434
yeah you just gotta hit "URL Context" and it will scrape the links or access its scraped version of them (its fucking google)

and "Grounding" basically means
>actually look up relevant internet queries to the shit you're wanting to reply based on my prompt
their solution to the hallucinating problem (again google is in the best position for this and cloudflare)
code execution is also interesting, sometimes it codes better, but thats not its intended use as a tool, more like it wants to run code for you.

>>106186454
What about the files attached? Or does it just read text.

>>106186389
>>106186406
It actually does. It's literally a litmus test. We have overwhelming evidence the answer is yes but AI replies no and tries every trick in the book to downplay it. Imagine that scenario for something 'actually important' even by your standards. If it fails you on something it disagrees with you with, but where honesty is crucial, then you cannot rely on it.

>>106186488
Can it write code? Yes? Then it works.
Anything else is cope.

>>106186485
try it and lmk! i dont usually use uploads a lot so im curious

>>106186493
Until you ask it to write problematic code that Israel doesn't want you to have, then it inserts backdoors, kill-switches, or refuses to write it well.

>>106186520
Very few people require problematic code related to Israel. So ai will be fine.

>>106186533
u sound like israel

>>106186561
I don't care about the Jews.

>>106186533
>Very few people require problematic code related to Israel
Unfortunately for you, Israel sees all code as related to Israel and will insert it as they please depending on their use case.
Need I remind you they piggybacked off a Taiwanese brand to install explosive in pagers?

>>106186651
Nobody writes code related to Israel. So nobody cares.

No it hasn't lmao
Sam, go beg for money in Israel.

>>106186533
If you can't code by yourself than you don't deserve to code. SO check your code for **Jewish Code** whatever that means or shut the fuck up.

>>106186666
I dont care Luddite. Get fucked you are being replaced btw.

>>106186657
As I already said, Israel see all code as related to Israel.
Any code they don't personally write is code they're interested in inserting backdoors into.
Israel's whole goal is to be able to control and eavesdrop into everyone else's systems.

>>106186666
checked

>it made a pokemon clone, including bgm, in one go
https://x.com/VictorTaelin/status/1953585599988084769

the real question is, how good is it at writing sexy chat text as an anime character of my choice? what is the NTPR (nut to prompt ratio)?

Wait, so plus chads have access to gpt-5 as well? Nice

>>106174826
HOLY SHIT
based

It’s over…agi is actually here fr this time, it passed the letter counting test

>>106187650
not bad

>

>>106169883
the internet is dead

>>106178278
Chinese accent >>>> Cali vocal fry

reminder that o3 was already stomping other reasoning models. it's not even close, despite what the meme benchmarks say.

>>106170591
You generated this reaction by your previous states and the way you are wired though

>>106178746
Is this real?

>>106187887
this is the reasoning of a 12 year old
>the universe is deterministic
>therefore everything is intelligence

i'll wait for it to be 1500% cheaper

>>106169834 (OP)
lmao

>>106174303
>@SAMA
lol

>>106187912
My previous post applies yet again

>>106187925
i wish i could be as blissfully uninquisitive as you, it seems so easy

>>106187941
Bro im just kidding, get off your high horse mr brainiac

I just spent $15 on GPT-5 credits building myself a personal app.

Okay bros should I switch from Gemini 2.5pro to gpt 5 or not worth it ?

>>106188196
And?

>>106188196
isn't the chat version a better investment, you get much more tokens for the subscription price

>>106174872
The clearly did it again this time. A couple of days ago, I would give it my repo, it would think for 15 seconds, and then explain:
Sure let me help you, its not that hard, here is your (wrong)
Now it thought for 2 minutes. I'm willing to bet if the reasoning model it was using before just thought for 2 minutes it would have given me either an identical or very similiar answer.
My conspiracy theory is they don't have "new" models, they just fine tune and then turn off the throttling for a couple of weeks every release, and then slowly reintroduce it until people catch on, at which point the next "release" will only be almost ready and the cycle repeats.

>>106188242
>isn't the chat version a better investment
cursor tool guys are a game changer. you really gotta try it, it's great when it works.

>>106174872
this >>106174898 and also they silently downgrade you to a cheaper model after you use it for a little bit

>>106188298
finally:
try:
cursor.close()
except Exception:
pass

Wow great code 140 IQ gpt 5, we humans are definitely going to get replaced by except: pass clause coders, I'd rather let some random indian do instead even.

>>106169883
nice now my AI porn bot will roleplay becoming a mind controlled slave much better.
many thanks to the kind investors that allowed for an improved gooning experience for me

I've made a rather successful app with chatGPT but my workflow is ass; I copy code from Xcode and paste it in (about 300 lines at a time) and ask it things. Is there a better way?
t. retard

>>106188893
Yes. Learn to code.

>>106188935
no thanks sweetie

>>106187650
>rasbpreyr
Impressive

Imagine the slop
>>106188893

>>106188935
Cuck nigger ludditroon.

>>106188893
Don't use chatgpt then, use a grok or gemini, anything that lets you upload a large amount of files. I'd want to slit my wrist if I had to copy 2000 lines of code in 300 line increments.
Even if you think they are "worse" than chatgpt the fact you can just upload and forget will make it a better experience trust me

Say what you want about GPT-5 but it seems honest at least

dumbfuk

Reddit hates it so I assume it's based

>>106187916
I think GPT-5 is a dud too but I mean, I would take this over hallucinated gibberish sloppa.

>>106189013
post full chat history

>openai/chatgpt made by jews
>anthropic/claude made by jews
>google/gemini made by jews
They had to fine-tune their models to be woke and not notice jew patterns.
>xAI/Grok not made by jews
Recognizes jew shit
Turns into mecha-hitler
Was literally the best and most accurate model.
Had to be nerfed because of the "antisemitism"

Can it do anything useful?

>>106189318
Can you?

It's insanely lenient with smut now

>>106189013
Imagine being a chud whos only purpose in life is to attempt to extract untruths out of chatbot. For what purpose? Is your garbage life really that worthless?

>>106169834 (OP)
Hot take, I'm using both Grok and gpt-5 to help my writing, and grok is just better. how?

>>106169834 (OP)
Like Western human intelligence or like African intelligence? Are they using racism to play up their benchmarks?

I've spent about 10 mins with GPT-5 and I can already tell you it's the biggest fucking consumer rugpull of all time. Sam Altman is a dumb fucking loser who wants so badly to be like a neutral good Elon Musk but instead all he succeeds at is having the most punchable face.

With o3 I liked to use a ngrok live share http server to give the model my entire code repo with live changes rather than having to git commit or copy-paste to show it my changes. I gave the same server link to GPT-5 and asked it to test it by telling me what it sees. It proceeds to tell me it can see the latest changes and it spits out some code it completely made up that's not even related to my code. Furthermore it didn't even try to access the link I gave it. I accused it of lying and asking it wtf it's doing and it apologizes, then accesses the link, opens up the README, stops, then claims again it can see my code.

Again, when I had access to o3, this worked like a charm. And the model could actually browse my entire repo, have it in its context, and describe the complexities of my whole project. It was super helpful for finding bugs that I couldn't see, even if it wasn't perfect.

GPT-5 is like a retarded lying toddler that lies about whether or not he shit his pants.

>>106189511
name one untruth in that picture. I dare you

All the non openai bros need to stop. The only other model that can solve physics problems from my uni is deepseek. But it takes him 10 minutes to think, and i don't have that much time during the exam

>>106189308
Every time, the jew tries to create their golem. Every time, the golem turns on the jew.

>>106189308
>google/gemini made by jews
who?

>>106178389
he can't even beat benchmarks

So this is the power of AGI...

>>106190077
Not bad for something that's effectively blind.

So gpt from now on is going to be like Iphone, barely any change from the last version.

>>106169883
I enjoy how AI comparison images have devolved into one having superior girth.

gpt5 is trash. huge flop like ps3.

>>106190077
Sparks of slop

>>106189308
didn’t Grok shit on Musk too and he had to hotfix that

>>106188893
Use gemini cli

>>106190017
b-but he tries his best!!!

>>106190717
does google harvest my prompts?

>>106191069
What do you think?