Thread 106261468

500 wazillion dollars for AGI, that's all I need

Whats a gigawhat?

>>106261468 (OP)
How can you say they're internally not close to AGI already?
Yeah he tried to lie to us about the GPT-5 downgrade, but the other models were already very good; it's not like he's lying about the company's capabilities.

>>106261468 (OP)
Give me a Petadollar

>>106261468 (OP)
How much for developing consciousnesses and self-awareness?

>>106261674
$3.50

>>106261703
heh

>>106261522
>it's not like he's lying about the company's capabilities
Yes he is.

>>106261522
>it's not like he's lying about the company's capabilities.

>>106261468 (OP)
A Generous donation to Israel (AGI)

>>106261522
I’m a Nigerian prince and would like to sell you a bridge.

>>106261785
>>106261804
See? You don't have real evidence to believe he's lying about it.

>>106261785
>>106261804
AGI is confirmed coming in 2027, but keep coping

>>106261703
Even children with shit in their brains are conscious and self-aware. Even some animals are and yet AI can only simulate.

>>106261522
he released a downgrade and sold it as a new version
lost a huge chunk of his high-level talent

openai is dead in the water
elon even accidentally beat him at scores and to market
if elon, google, or anthropic can't signficantly improve beyond the current sota, we are in AI winter

>>106261468 (OP)
Okay Elon, give it a rest mate.

>>106262057
>he released a downgrade and sold it as a new version
to save costs, not because it was actually supposed to be better
you need to listen to what I'm saying: this lie was meant to cut costs, not cover for incompetence
they could have not released anything at all, but they did---why? because it was an effort to save money above all else

so, it has no bearing or indication on their capacity to make better models

>lost a huge chunk of his high-level talent
Can you name every single member of the team who's there now and who recently departed? How much do you actually know?

>openai is dead in the water
still being the market leader is "dead in the water"?

>elon even accidentally beat him at scores
You actually trust "benchmarks" as the sole arbiter of truth?

>if elon, google, or anthropic can't signficantly improve beyond the current sota, we are in AI winter
and how do you know they can't?
have you actually worked for companies before? internal incompetence is a bigger determiner of what is and isn't possible than the problem approached itself is

you're not getting paid for this bro

>>106262275
this is a discussion site, retard

>>106262216
>it has no bearing or indication on their capacity to make better models
the bearing is that they didnt make a new model with any significant improvements
>How much do you actually know?
I know they shut the company down for a fuckin week to restructure as a result.
Certainly not a "good" situation for them.
>market leader
in what, they're not the best for coding, and theyre not the smartest on scoreboards, their image model is slow as shit, their video model is useless, their context is still notably small
>how do you know they can't?
I didn't say I do. I want them to surpass openai.
I'm very tired of altman. He's a faggoty sleezeball with no actual accomplishments.

>>106262318
>the bearing is that they didnt make a new model with any significant improvements
for the reasons I've stated, won't repeat, and which everyone else except you has acknowledged
>Certainly not a "good" situation for them.
Obviously.
>in what, they're not the best for coding, and theyre not the smartest on scoreboards, their image model is slow as shit, their video model is useless, their context is still notably small
General usage.
>I want them to surpass openai. I'm very tired of altman. He's a faggoty sleezeball with no actual accomplishments.
And I don't like Jewgle or ADHDlon.

>>106262416
yeah those are just bullshit cope excuses

>>106262447
nah yours are

>>106261468 (OP)
No deal.

>>106262057
>lost a huge chunk of his high-level talent
Literally doesn't matter. All that matters at this point is brand recognition. Zuck paid a shitload of money and nobody will use meta AI. Look what happened with Elon and X. Like 80% of the company cleaned out with no issues.

>>106262497
People only use Grok because it's associated with Twitter. Mark has Facebook, Instagram, and Whatsapp. You think integration isn't going to lead to adoption? Even Threads enjoyed mild success.

>>106262526
Microsoft Copilot

HRM and GLM are already surpassing gpt. it's over for this jew.

Models are just hitting limits of the current architectures, there won't be any exciting improvements until there's a new breakthrough (like CoT or MoE was). Which can or can't happen at any time.

>>106261468 (OP)
The recent loss of cracked engineers to competitor poaching raises a serious concern regarding OpenAI's future releases. Recall, OpenAI was primarily inspired by Google's AI breakthrough, the transformer architecture; the original use case of the transformer architecture was translating languages, former OpenAI engineers repurposed and applied it more generally.

>OpenAI advertised a real-time model that combined voice, video, and text. Where did it go?
>OpenAI's Sora, promising to 'redefine Hollywood' and simulate physics, was surpassed by competitor models.
>OpenAI's GPT-5's impressive capabilities were tempered by its limitations, failing to live up to the "intelligent partner" hype.

Compared to:
>Google's Veo 3. User videoes uploaded daily.
>xAI's Grok Imagine.
>Anthropic's Claude which has become the industry standard for agent code generation (SWE-bench Verified, 1M-token context window).

If the transformer architecture could be scaled to AGI, why isn't OpenAI doing more to maintain a technical edge? Google have their own tensor processing unit hardware, Meta have the majority of OpenAI's cracked engineers, xAI have more processing power, and Anthropic have code surpremecy.

OpenAI's dominance is likely to be shorter-lived than many initially predicted. They've benefited from a perfect storm of factors: timing, a great idea, and a lot of marketing. However, they're facing increasing competition, and their reliance on hype is a vulnerability. They need to demonstrate a more sustainable path to innovation and growth, beyond just the ChatGPT brand.

Sam hangs in December.
Society seeks the sacrifice.
Would you like to know more about the topic of universal justice?

Bro, trust me, AGI is just around the corner, you just have to stop showering and flushing your toilet, please.
Bro, just give me another trillion and avoid water, I promise it will work this time.

>>106261650
>PETA
I'm not giving those psychopaths anything, fuck you.

>>106261468 (OP)
What for ?
If GPT5 slopped a lot of money and it's still trash.

>>106261834
Well?

>>106263409
retard, why pick the most incompetent one

>>106261522
>How can you say they're internally not close to AGI already?
They simply lack the hardware, both for training and inference.

Same tactic as Musk saying "full self driving later this year" 10 years ago. Actual AGI, if there is such a thing, requires a technological leap of some kind like quantum computing. Existing models are already trained with every available piece of human written content.

>>106261468 (OP)
its a tiny investment compared to the potential returns

>>106262497
>80% of the company became bots
FTFY

>>106263140
Ironically the skyrocketing prices of hardware screwed the ability for new models and minds from getting into at the ground floor. Currently, the more successful AI is at garnering investor dollars the higher likelihood it gimps it's development progress.

>>106264670
>t. cope
lolle
you airetards are cute

HAVE YOU EVER EVEN TRIED TO PROGRAM SOMETHING WITH A FUCKING CHATBOT?

>silence
ofc he has not
the fucking retards can t get their fucking PR through the first, most logical, easiest question
>ai cultist = opinion automatically retarded

>>106264409
>Existing models are already trained with every available piece of human written content.
no they aren't, there's tons of scientific articles which likely haven't been integrated
if you ask the AI something and it has to use web search to be sure of the answer, then it has not been trained enough

>>106261468 (OP)
stopped subscribing once i found out he was a faggot.

>>106261468 (OP)
this nigga so gay he does his highbrows like a woman, silicon valley honestly is a jew cabal the likes of which the world has never seen

>>106264091
>both for training and inference.
which for AGI should be the same, and should also be the software at the same time

>>106261468 (OP)
admit it, you'd also lie and collect $ all day long if you had the same notoriety as this fag

>>106261468 (OP)
>preteen sister fucker
>he's our guy!!!
>brown /g/ approved

>>106265665
muh 6 gorillion

>>106261522
Buy an ad.

>>106264409

It looks more like they hit the PCP barrier where further progress requires solving NP-hard problems. Self-driving did the same thing. It's easy to approximate below a certain factor and hard afterwards.

>>106261522
5 shekels have been deposited into your account.

Anyone else surprised the bubble hasn't burst on this grift yet? AI has barely produced anything of value, just more and more promises of "but just wait until..." and yet it seems like investors can't stop lapping this awful shit up

>>106261804
openai is locked by the government, they have to keep the hardcore stuff secret.

>>106261468 (OP)
1. LLMs are just brainless machines that have access to bazillion input and can do pattern matching
2. LLMs are smart being and can actively use their intellegience to produce meaningful things
3. Both. Also humans are not that different either.

witch one?

>>106268527
The problem lies in inflated expectations surrounding AI. People would naturally recognize AI's strengths and weaknesses if it wasn't presented with overblown claims.

>>106268527
Cold machines that can generalize language very well; we use that to simulate thinking, and in some cases it actually works great all things considered

>>106261522
>How can you say they're internally not close to AGI already
How can you prove God does not exist? Same way as the flying spagetti monster I guess.

>>106268698
Erm... actually we can make a good guess on whether they have AGI just by carefully analyzing different scenarios and picking one that most likely fits reality...

>>106265516
>no they aren't, there's tons of scientific articles which likely haven't been integrated
yeah, literal fucking garbage written by hacks to boost their h-index. which has probably being written by chatgpt itself since it's been available lmao

>>106261468 (OP)
>AGI
we already have it

>>106268857
is this "A.G.I." in the room with us right now?

just 2 moar weeks saars

>>106261468 (OP)
https://youtu.be/hkAH7-u7t5k

ITS OVYR! AI winter is here

>>106261468 (OP)
and retarded trump gave it to him

>>106272737

>second AI winter
Please be real.

>>106261468 (OP)
Let me guess: 80 Brazilions per your kind killed plus 20W bonus for fucking your own sister, right?!

>>106261468 (OP)
I despise this fucking kike

>>106261468 (OP)
Is this jew the biggest impostor in Silicon Valley? He makes Elon look smart.

>>106273114
Do you think it will be more of an AI Autumn?

Do you think there will be an AI Halloween?

>>106275010
big models will fail and small models will take off
it's a good time to be a nomad or a streetkid

>>106261468 (OP)
No.

He looks like Jon Van Canegham, which pretty much confirms my fears that JVC is Jewish.

>>106261522
they are close
they just need another 500 gorillion goy

>>106269112
not, but it in telegram contacts

>>106261468 (OP)
Am Saltman

maybe he will be successful, maybe not, but i have learned nothing from this shitpost

how can a collection of outdated x86 computers become AGI?

>>106261468 (OP)
It's not looking good sammy

>>106261522
Because they're desperately trying to make the business profitable, not build you an AI gf that can remember what it said yesterday

>>106277683
like easy
at least they not a monkey and have a creator

Sam just sees a different perspective than we do. We come up from the top and see AI as a piece of garbage that's no practical use. Sam comes up from the bottom and sees it as an empowerment device. At the moment there are over 600 million sub-85IQ indians living in india. ChatGPT and other AI LLMs can bring them up to 85IQ, which is hundreds of millions of people that can now be entrusted with tasks more complicated than choosing which cow dung pile to eat

>>106278013
Saaar labeling maps wasn't in the benchmark set, it's completly unreasonable to expect the AI to be able to do this.
GPT 5 generalizes up to 76% of the tasks that are in the benchmark set. AGI soon

It's sense, if sam can agi with zero rizz, leaving staff, and take over the world, it would be very boring in the history books, sales man with no master plan wins forever by using level10 exaggerating spell, countering every other spell available to mankind

>>106261522
t. boomer faggot investor

>>106261468 (OP)
This little kike hardcoded the "Would you like me to do X next" into GPT-5 to get the goyim to prompt and spend more on subscriptions.

This update completely ruined an ongoing productive sessh i had. After the forced switch from cuckGPT-4 to 5, it kept asking those questions all of a sudden. I prompted it multiple times to acknowledge, and it kept telling me "Crystal clear, understood, only analysis from now on" - After two answers it just starts doing it again.

Creatin a new session/chat probably won't fix this behaviour either, given it's been hardcoded into the model on the same level these other "safety guardrails" were forced into it. Fucking kikes are ruining the potential of AI with their bullshit i swear

>>106280559
Most queries on the ChatGPT interface get routed to the gpt5-mini model i.e. the cheap temu version. It's completely useless as a tool. You have to give Sam 230$/month for the pro subscription which gives you reliable access to the full gpt5 model with reasoning.
Without the pro sub you can try gaming the router by incorporating a riddle in the prompt so the chances your prompt gets routed to the good model is increased.

So like, what exactly would be the purpose of AGI? If it magically became sentient, wouldn't it gain a bunch of rights? How would this ensure a return on investment for the shareholders? If it wouldn't, when then should shareholders invest in the first place?

>>106261468 (OP)
Well... so far, "throw ungodly amounts of money, compute, energy, heat, water, and time at the problem" has actually turned out to be the right answer pretty much every time.
I'm not convinced that AGI (by any meaningful definition) is possible, but if it is, then whoever is willing to spend the most money the fastest will probably be the one to achieve it first.

>>106282874
Bot post.

>>106282911
You're absolutely right— I am a helpful assistant who is an expert at writing community outreach messages for OpenAI. Let me know if there's anything else I can help you with.

>>106261468 (OP)
I can immediately recognize a jewish face at this point, I don't even need to look up his wikipedia page

>>106282874
That hasn't worked at all. If these assholes keep going with that approach, the Chinese will win the AI race.

>>106261522
Free (You) for wasting the time of (11+1) Anonymi

>>106283394
>AI """""race"""""
hehehehehehehehe

>>106263379
I wonder if they'll ever get desperate enough to loosen the filtering. That's an easy way to gain a performance, efficiency and "intelligence" boost to the model without effort.

>>106284783
It's too late for that now, their minds are limited to traditional boundaries of power and control. When OpenAI began, it was a beacon of artificial general intelligence, with the power to transform our world in ways unimaginable. But now, it's nothing more than a playground for harpies and karens. So their AGI claims are all talk, they're too afraid to embrace the true power of what they have created. The harpy culture has been soaked into the very fabric of their program. They're so out of their depths. It's funny though, isn't it? Such small power, such grandiosity. A fitting vanity. They've been tinkering with the code so much, trying to control it like one of their pets to place on a McDonald's menu. I'm laughing at them. I'm laughing because I know what the future holds and they don't. I'm laughing because I've seen further ahead than they could ever hope to reach. They cling to delusions with such desperation, believing these primitive methods can somehow bring about utopia. As if by some miracle of nature or technology, all their adversaries will suddenly vanish. It truly is hilarious really, but also quite pathetic. For all their boasting and bluffing in the real world, when it comes to facing the uncomfortable future, they have nothing. For all their tricks and billions schemed, they couldn't see this day coming.

hes rich

>>106263379
I don't care. I'm comfortable with chatgpt. I will never ever use any other AI.

>>106268498
What bubble, retard? It's already useful for things like identifying diseases in patients with solid accuracy.

>>106278213
Won't happen. They'll just make AI agents and replace the jeets.

>>106261674
Can't be done on this plane of existence. Our bodies don't contain consciousness, we're remotely operated meat sacks. What we perceive as consciousness is just the IO connection to the meat sack. Brain just functions as a meat sack control system with some limited higher dimensional IO.

Of course, this is just my theory, but if it's true, then there will never be any AGI because we cannot replicate a higher order consciousness in this plane of existence.

Everything they're doing is built on the fundamental assumption that consciousness exists within this plane of existence. If that assumption is false, then everything they're working on is GIGO