>>109121732
Sell your knowledge to china

is this a cris thread?

>>109121732
Post about it on a public forum

>>109121741
But I hate communists way more than I love money. Waaaay more.

>>109121732
>inb4 muh world models

did you start believing your ai girlfriend is real?

File: dr-makise-kurisu-attempti(...).jpg (110 KB, 640x459)

110 KB JPG

>>109121744
There is definitely need in more Kurisu. If only I had an assistant like that...
>>109121762
AI gfs are dumb hallucinating schizos, not into that.
>>109121756
If you are talking about simulations that require infinite amount of resources and other singularity homo nonsense, it's far from it.
Nothing magical though. There's a reason I call it AGI, although I hate the term, invented by grifters. It's definitely not super AI, because that one is impossible to create.

Can't say much more, because it's really simple.

>>109121819
you solved AGI yet you dont know what a world model is?

>>109121858
Yes.
Well, I've seen how they train self driving cars and walking robots, but still not sure what you are referring to specifically. I thought that you thought that I thought...

File: makise.jpg (58 KB, 603x631)

58 KB JPG

I know what you're thinking, anon. That I'm just like Sam Altman. A fucking scammer. But it's not like that. I'm not trying to sell anything here or get you to subscribe for something.
Maybe I am mining for ideas, that's true. But at least not hiding it.
>>109121746
A place where people share ideas seems like a good fit.

>>109121732
let me guess
>there is no such thing as general intelligence
>there is only generally applicable narrow intelligence
>the brain is itself bootstrapped and narrow
yeah, we know, we figured this out a while ago

File: smug assistant.jpg (69 KB, 683x384)

69 KB JPG

>>109122930
> there is no such thing as general intelligence
No there is. But it has to be combined with specialized intelligence to be usable. That much obvious I think. Clearly people understand this, there are these generalized orchestration models, like the one Japs presented recently. You can see that OpenAI is doing the same, as that "leak" of information from their codebase suggests. The "ultra" mode would utilise orchestration layer similar to Sakana Fugu.

Anyways, all of the big AI companies are doing it wrong. I don't know why though. I was thinking maybe it's a psyop to make sure most countries in the world are behind USA. Either they are onto something and that is the reason for such a crazy amount of datacenters. Or they are doing the usual, betting on some chances, laundering money, securing land and property etc. I don't know which one it is.

What I know is how to make something that would be much more productive than a single person and would be able to solve problems not only a single person struggles with, but those that require very large amounts of people working together. That the level of a large corporation or even a state actor.

>>109121732
>What do?
take meds

>>109121750
Good thing they're national socialists then, just like the Germans were.
Sell it to Japan

>>109123493
Have you tested it, or are you just blowing a theory out of your ass? Because as an ideas guy trying to conceptualize a solution to the problem everyone is working on, chances are you're late to the party.

>>109123493
do you need gpus to run it

File: russia.jpg (126 KB, 1280x720)

126 KB JPG

>>109123927
Pretty sure I couldn't sell it to you, not even for $5. It has as much value as Altman's promises of AGI, why would you believe me, not him?
That's the problem. And a deadlock. To have credibility, it should be implemented and demostrated. But to avoid it being snatched out by OpenAI and alikes, it should not be demonstrated and should not have visible credibility.
This thread is pretty much about trying to resolve this problem. It can be applied in many different fields maybe. If solved.
>>109123954
I thought I am late last year. But guess what, nothing happened. Nobody tried that. Nobody spoke of it. "It" being a very simple and obvious thing, that's the shocking part for me.
Previously things I was talking about to people of random AI harness-related forums materialized as real killer features included in premium plans.
None of those ideas were my own, if anything. Don't get the wrong impression, I'm not delusional, just saying that I've seen AI companies monetise those things pretty quick. But only if they know it exists and proven to work. They minimize RnD spending otherwise.
>>109123989
Pretty sure I can run it on rented compute alright.

>>109121732
I genuinely have a remarkable idea about AGI, it will completely shatter the way you think about the problem but I don't know who to go to. Definitely not OAI or Anthropic because frankly none of their people seem to be able to grasp the scope of the problem. What do?

>>109124096
Could be thst they've just considered it already and it doesn't work, which is why it isn't talked about. I had a potential idea about RAG that no one seemed to be considering, turned out they already had ages ago and that was why it was buried.

Your best bet is to make a proper documented paper on the concept, then either patent it or (if you rightfully don't trust your government) approach a startup AI company, DEMONSTRATE that your shit works, and then sell it for share in the business.

File: do it for her.jpg (50 KB, 680x511)

50 KB JPG

>>109124099
Nice mocking attempt, but it's a bit off.
They do understand the problem, most of people ITT do. But they all try to solve it in an overcomplicated way. And they all copy each other. They are afraid to waste money.
Probably because their sutiation with money is not looking good.
I imagine if I casually ask Altman to lend me $10, he will politely decline the request and then might cry a little, LMAO.
>>109124140
>Could be thst they've just considered it already and it doesn't work, which is why it isn't talked about
Possibly, but what about random people, who actually try out crazy stuff and innovate a lot with what they have access to in their free time? They are sleeping on this also.
> Your best bet is to make a proper documented paper on the concept, then either patent it or (if you rightfully don't trust your government) approach a startup AI company, DEMONSTRATE that your shit works, and then sell it for share in the business.
Why do you believe that it's the best bet? Nothing is wrong with it, but sounds a default option. Can the default option be ever considered best in such matters?

>>109124096
>Pretty sure I can run it on rented compute alright.
lmao, then why don't you solve the arc agi problems, claude 4.8 got a measly 1.5%, i'm sure yours can do better

>>109124096
>>109124099
Just use your ai to make a good product/breaktrough discovery, popularity is as an ensurance against idea theft

i'll buy the idea from you for $10k, interested?

>>109121732
They can't kill you if you set up a dead man's switch and kill yourself. Foolproof.

>>109124252
Sure, I can also license it to you.

>>109124192
What do you actually want out of this? If it's recognition, you could probably get it by just dropping the idea here. "4chan solves agi" is the kind of title normies lap up.
>sounds a default option
The default is just coming out with it, which like you said will then get stolen by corpos. Getting a small company on your side (which will then theoretically become a big company, if your idea is as good as you say), is insurance against that.

Speaking of which, can you actually define AGI

>>109124252
How do we proceed? Do I email you my cvv code?

No deal. Bare minimum is access for me to latest compute in your possession, limit beling 1 GWh of electricity per year, for 10 years.
Unless all compute you have is 1080ti, that's a different story.

imagine larping you have agi on 4chen instead of spending time making an mvp, you don't even believe yourself do you

File: 4.jpg (250 KB, 847x632)

250 KB JPG

>>109124280
> can you actually define AGI
Sure. Sam Altman can't do that, but I can. AGI can solve same problems a human can. Or same problems that require a group. Or same problems that require an enterprise etc. It's killer feature is scalability. On smol scale it is not capable of getting anything done from start to finished product, just like modern AI.
And it is still just a program on a computer, information in, information out. No skynet bull shit, no cure for all diseases as that is very closely tied to real physical world. Can't be more specific now, but hopefully distinction is clear.
>>109124327
There are many things I do for the lulz. Handing over tech that half of the world is hellbent on inventing is not one. Not because of greed, but more like because it will probably end up in 'Murica. And modern 'Muriccans are fags, they will leak in to commies. I hate commies so much it's unreal.

Japan and France currently sound like good options. France being a strong one, since they already have a player. Le Chatoon Fat meme being manifested in reality if the kind of lulz I'm into.
No idea what Japs are doing right now, besides the Sakana Fugu thing.

>>109124378
I am not telling you to hand over anything pseud, i am saying to make something with it that will make people take you seriously, you just sound like a naive teenager that has just discovered the taste of beer

>>109124427
>make something with it that will make people take you seriously
Like benchmaxing? I'm not familiar with that, you think a random can use it through API or something and then have results published? Or is it backwards, like they use my API endpooints to solve problems and then log results?

it's just continuous learning... once it's fully solved you can teach clankers to do anything you want

>>109124449
You don't even know what you should be making lmao, if this is indeed agi, it should be cheaper to make anything that current llms or physical ai are making. Your product should be instantly cheaper than anything on the market. You have no idea what you are doing do you? You talked to some llm that convinced you, is this your first case of ai psychosis?

Do a satoshi nakamoto if it is that good

File: 1773497872473.gif (1.89 MB, 498x498)

1.89 MB GIF

talk is cheap
if your ideas are actually any good, write a paper and post it in arxiv. if you want to have anything to show for it, implement it

>>109124463
>Your product should be instantly cheaper than anything on the market
What they ask for their product depends on their planned RoI. And most players on the market already have a spot under the Sun and are able to plan for something in general. So it's not that simple. Go try to make cheaper and better search engine than goggle. Even if you will, people will use goggle search because everyone uses it already, they already have the money to bribe everyone into making it the default absolutely everywhere.
>You have no idea what you are doing do you?
You are absolutely right, but that proves nothing.
>is this your first case of ai psychosis?
No such thing. Shills bots just hallucinate from time to time, people call it phychosis and laugh, but it's not funny to get spammed by that shit.
There's no psychosis ITT, just a calm discussion.

So? Which benchmark would you suggest? It should be something that would leave no doubt that AGI is actually real. But not to be confused with strong AI tasks, like "solve 100 unsolvable math problems".
Running this shit would cost me a shit tonn of money, just to benchmark it.

>>109124519
I've already given you arc agi 3, but all you do is talk. You obviously have nothing since you don't even know what benchmark to try first.

File: kona.png (185 KB, 450x338)

185 KB PNG

>I have figured out AGI [...]
>What do?
If truly have figured out AGI just ask your damned robot what to do. It if it's so smart it will be able to tell you the correct course of action.
Unless you "figured out AGI" solely in your head which basically means you're just speculating and haven't actually solved shit since you haven't even tested it

>>109124529
>you don't even know what benchmark to try first
That's the neat part. I was not bothered with all this garbage over the years.

Just checked it out. It's some kind of IQ test for retarded people. You joking?

>>>109124378
AGI can solve same problems a human can. Or same problems that require a group.
So why are you asking for help on a senegalese basket weaving instead of asking your AGI what to do

File: MADDO SCIENTISTO.jpg (120 KB, 1366x768)

120 KB JPG

>>109124543
Yeah, I'm afraid OP just comes across as an ideas guy (pic related). He might genuinely have a good idea but it might be another Future Gadget #003.

OP needs a Daru, but that's hard to come by here. I mean I have the hardware and I'm down for some tests, but I also have massive trust issues so I'm not running some rando's code.

>>109124586
He doesn't have a setup

>>109124567
Lmao, you are a pseud, let this be a lesson to anyone who engages with this brain-fried retard.

>>109124588
you could run it on hypervisor or some shit but i doubt op even has any code to send you in any case
>inb4 no no no i have agi i just cant trust anyone i cant show it to anyone
hahaha

>>109124592
So does OP not a have a job or anything? No source of income to pay someone else to lend him some compute? Not even access to some college cluster to test shit in?

>>109124611
He never claimed to have agi, just that he "figured it out". Hence, ideas guy.
>>109124630
Basically, yeah. OP, before you start to even think about sharing this with a company, you need to make sure it works. That's where your efforts should be focused on. Get a working prototype, then you can start showing it off to potentials.

(or if you really want to fuck over the corpos, open source it so no one can claim ownership).

File: Steins;Gate episode 03 - (...).png (1.8 MB, 1920x1080)

1.8 MB PNG

>>109121732

You should do what the characters of “Steins;Gate” would have done with the time machine.

Do not publish anything, do not build anything because it would end the world as we know it and also would start some WW3.

So OP ya a neet or a college kid? You ever gone to college? This is a serious question

File: 1781951391308.jpg (48 KB, 470x652)

48 KB JPG

Nuke India.

>>109124593
Altman is a pseud, sure, since he defined what he believes AGI is, but never delivered. His gay geepeety cannot pass test he promised it would pass.
I never promised real time vision and puzzle solving model, that's not what AGI is for, not per my definition.
>>109124630
I do, but no professional ML background. Hence questions about current benchmarx. Last time I checked, "humanity last exam something" was the hardest.

>>109124650
>That's where your efforts should be focused on. Get a working prototype, then you can start showing it off to potentials.
You're almost right. Last half a year or so I spent on security and homelab maxxing. Because it is not really possible to get anything done without getting robbed these days. Loonix may be pwned any second, microslop screenshots everything, apple scans hard drives. And those are not jokes anymore, but daily occurrence.
About to roll out DPI on a custom router and IoC provisioning system on proxmox.
>>109124659
OP is not like Yudkovskyi, don't worry about it.

>>109121732
Polymarket or something

File: 4.jpg (111 KB, 848x1272)

111 KB JPG

>>109124722
Indeed.
>>109124732
Polymarket what? Last time I checked these casino type facilities only existed as a mechanism to bribe big shots by "losing money", while casino usually belongs to family member of bribe recepient or something like that. A scheme with an elaborate front basically.
So even if there was a good way to apply AGI to predict something, they would probably pull out some exceptions from their asses and refuse to follow their own rules, ot avoid payouts. And ban me ofc. Like it happens in regular casinos.

>>109124694
Your definition was that it could do anything that a human/may humans could accomplish.
A human can see things in real time and solve babby puzzles.

>>109124732
But I like the idea. Just quitely making a lot of money by releasing products, doing business and such. Not casinos though, don't get me started on this, after all this is tech board, no place for antisemitism here.

>I do, but no professional ML background.
What do you mean? You haven't worked with ML, alright, but have you done any serious academic research on it? Taken courses, published any papers?

>>109124751
>see things in real time
That's very specific. And hey, at least I'm not lying that it can do it, when it clearly cannot.
Altman does that, by the way. I'm still better than him.

>>109124762
It is something a human, no - nearly EVERY human - can do.
How dare you call yourself better than Altman when you're lying about what you have?

i agree with OP, real machine intelligence is actually trivial, it's just also fundamentally schizophrenic

Op, post the idea here, cuz i wanna know

File: 4.jpg (98 KB, 618x808)

98 KB JPG

>>109124788
Ah shit, you got me.
You got me good, I am actually as good as Sam Altman, not better than him.
Technically I lied, but it was not intentional, just bad choice of words. Maybe I should just pick a better term for this. But anon it's honestly hard to do, the solution to many currently unsolved problems is trivial to such an extent that I can't give it a simple name, it will give out my idea.
So it's basically AGI, but it does not include vision or real time capabilities. Meaning, it may take way too long to solve a problem compared to what it would take a human. It also may be way more expensive than what it costs to employ a human.
But it's scalable, which means you can have more "workforce" than you would by simply hiring humans. There are not as many competent people to hire, while there's almost no limit to how much compute you can throw at a problem.

So it's not fater at solving simple problems, neither it is cheaper. But it can do what an enterprise can do. Without having to hire 10000 people, keep them in check, make sure they're happy, make sure they're busy etc.
The reason it is like that is because I was trying to solve a problem of that scale, not to do some dumb riddles for children.
So how do you wanna call something like that?

>>109124788
Not really, no.

>>109125076
Misquote >>109125055

>>109125055
So what's the problem, you lack the needed compute to run it?

>>109121732
can your AGI give you a blowjob? if not then it's not AGI.

>>109125055
Sounds like you're just suggesting to add more agents, lol

>>109125055
an agi should be able to do both children's puzzles faster and better than any human, child or adult and still be able to scale to do the work of 10000 people, it just sounds like you don't understand what you are building or how it compares to current solutions so you slap an agi sticker on it and hope it fools everyone, stop typing paragraphs and start building proofs

File: kurisu.jpg (12 KB, 474x248)

12 KB JPG

>>109121744
Every thread is a kurisu thread

>>109125092
No, it's the AI companies.
First of all, they would make situation with hardware even worse. Because currently there is limit to how useful AI can be until it needs supervision. My approach basically removes this limit. They would ruin the world economy faster than they would hit a bottleneck, like some sort of hardware bandwith limitation.
Second of all, they will leak it communist China, like they always do.
>>109125136
Agents don't really scale and can't really do anything complex without supervision. You had 1 problematic semi functional model. You multiplied it, got 100 problematic semi functional models burning tokens without supervision.
Dead end idea.
>>109125160
>agi should be able to do both children's puzzles faster and better than any human
You are talking about magical super AI. It is impossible to create, even Sam Altman couldn't deny it, he had to fall back to "AGI" term.
>be able to scale to do the work of 10000 people
By hiring 10000 people. Instead of that you can edit config file to utilize more compute. It's different kind of scaling. And again, current AI does not scale. Not in a sense like "more AI models working on same problem = better", go ask around all the people who tried using agents. Ask them how they are doing, financially especially. And if they solved any problems they were trying to solve that way.

>>109125243
So you don't lack the computer. Are you running it right now?

>>109125243
>needs to edit config file to tell his ai how to scale
ngmi, maybe go back to being a steins gate fag, you are more authentic that way than what you are trying to larp as here

>>109125281
AI gooners are built different though. They are gangbanging deepseek as we speak.

>>109125055
>trivial
>does not include
>real time capabilities
>may take way too long to solve a problem compared to what it would take a human
>may be way more expensive than what it costs to employ a human
>it's scalable
>but also more expensive and slower than a human
>there's almost no limit to how much compute you can throw at a problem
The "limit" is how expensive and slow it is compared to just hiring a human. The reason why nobody has your "AGI" is because brute forcing problems isn't a viable strategy.
>the solution to many currently unsolved problems is trivial to such an extent that I can't give it a simple name
It's either brute-force (not viable) or just ignore it (also not viable/desirable for companies).

>>109125575
the reason why nobody has OP's "AGI" is because it doesn't exist

>>109125575
>The "limit" is how expensive and slow it is compared to just hiring a human
True, but business maintenance is a hugely expensive and exhausting thing. You wouldn't start one and shut it down during the course of the following week? But being able to do that without hiring office full of people gotta be worth something, right?
> It's either brute-force (not viable) or just ignore it (also not viable/desirable for companies).
Nah, nobody does that, nobody even speaks about it. I'm telling you I thought by now it would be commercially available and figured I'll just wait 'til it becomes a thing, rather than wasting time on doing it myself.
Because previously most of my wet dreams were implemented like that. Agents, tools calling, specialized models for editing files ("fill in the middle" kind or whatever). AI companies just did that and I can use that for a dime if I need to. Spend 0 seconds of my life implementing that. Ideal.
But this "AGI" stuff? There are companies walking straight past this shit as we speak. Yesterdays news from OpenAI and Jap's model too, they are literally going straight into a dead end.
I know it's a dead end, because Cursor and Windsurf and alike have such models already and it's nothing new for them and they know it's limits. Sakana Fugu and OpenAI will hit it with their heads once again.

>>109121732
Have you implemented anything or is this some idea in your head?

>>109125848
Fine then. Drop the idea here, and I'll implement it for you.

File: 1776468055601600.gif (93 KB, 200x200)

93 KB GIF

>>109121732
>What do?
Use it to cure your autistic symptoms such as being a tranny and being into tranime (the true mark of a midwit).

>>109125989
Hi Sam, how are you doing? Not well, huh?

is your system deterministic or probabilistic, let me guess, you'll type another paragraph explaining nothing and going on a tangent about altman?

>>109126128
The way you democratize AGI is by releasing it instead of waiting for this "trivial idea and implementation" to be discovered by these billion dollar companies. Worst case you allow someone to break into the AI space, likely case (99.99999%) your idea was discarded because it doesn't work or is not computationally viable (slower and more expensive than just getting a team of humans).
Or you just implement it yourself, if it's so trivial and easy you should be able to create a scaled down version that can solve toy problems (that ARC AGI test or whatever the fuck it was).

File: 138203569_p0_master1200.jpg (964 KB, 900x1200)

964 KB JPG

>>109125955
There's not much to do really. It's like building something from off-the-shelf components. The only new thing is a specific combination of already existing tech.
>>109126147
That sounds a bit vague. If I understand you correctly, it's probablistic. Definitely not a magical math formula impremented in esoteric LISP dialect that makes computers sentient.
> you'll type another paragraph explaining nothing and going on a tangent about altman?
Anons also mostly try to ignore the purpose of the thread as well, talk about details of a thing they've never seen and never gib advice. That's ok.
>>109126172
>The way you democratize AGI is by releasing it
They literally have laws lobbied into existance that prevent competition to big AI companies. Go ask around in AI gooners thread, they are literally banned from raping computers. I don't know why, perhaps they prefer it when pedos hunt for kids. Either way it's a fact that someone made something like that illegal and just like that they'll figure out how to make "too powerful" tech just as illegal.
What do you think will happen if AI giants turn obsolete overnight?
Definitely not democratization of AGI.
> Or you just implement it yourself
On it. It's not implemented yet not because it's hard or a skill issue. I thought someone else will do it. Last year.

>>109126261
>laws
Oh no the laws are going to put the genie back into the bottle!!! They're going to wipe the information remotely from everyone's devices, and use the Men in Black forget machine to stop people from remembering how to do it!

>>109126261
>There's not much to do really
I didn't ask you regarding the amount of work needed
I asked you if you have done the work. Anything at all
Do you have anything to show for it?

why does this board have so many schizo threads?

>>109126261
>Anons also mostly try to ignore the purpose of the thread as well, talk about details of a thing they've never seen and never gib advice. That's ok.
You got your advice already, implement the fucking thing before you talk

>>109121732
>pretend I don't exist. If I insist, they'll make sure I don't exist.
Give it to me. When I use the AGI to exterminate humanity I will create great statues to honor your contribution and ensure that your mark lasts forever.

>>109126358
Hey Bill! How is your farming business?

File: nav.png (158 KB, 360x360)

158 KB PNG

>>109124519
>So? Which benchmark would you suggest? It should be something that would leave no doubt that AGI is actually real.
Here's a good benchmark: ask your "AGI" how to solve the exact problems you've outlined in the thread. If it can solve them, it's sufficiently intelliget. If it fails to solve them, it makes no difference that if fails; your plans of gaining credit for solving AGI aren't foiled, all it means is that you never had it in the first place, and no one will care for it at all.

You won't do any of that though, because you're just some neet with delusions of grandeur, much like many others that pop up on the board now and then, and you can't come up with anything other than excuses.

>>109126374
It's going great. I can spread my ticks around and people defend it by saying they're "sterile." We've started experimenting with screw worms, and people are actually blaming Trump instead me!

Have you asked the chatbots why your genius idea hasn't been implemented yet? You might get some answers that way.

File: 3825588072.jpg (735 KB, 2880x1800)

735 KB JPG

>>109126602

>>109126589
Pretty sure I did. It was a research mode as well. With active sub and top model.
>>109126391
> ask your "AGI" how to solve the exact problems you've outlined in the thread
How do I BTFO OpenAI and Elon Musk?
> If it can solve them, it's sufficiently intelligent
No shit.
Problem being how expensive it would be, also how tightly couples it would be with processes in a physical world, how dependent would it be on sufficient reaction speed to real time events etc.
Anon, there's an important distinction between problems that are computationally solvable and unsolvable ones. What you are propossing is essentially predicting the outcome of some kind of corporate war between Scam Altman and me. Although we are equals, I don't have nearly as much resources, so it would result in "AGI" infinitely iterating trough ideas on how to do that and throwing each of them out of the windows. You can tell without "AGI" that chances are slim. No matter how I apply it, a nerd with a computer cannot wage war on transnational corporations, which throw tens of billions of dollars at their problems left and right 4 years straight.
They can literally kill me by dropping like 3 metric tonns of money (all 100 dollar bills) from a helicopter on my house and then setting it on fire. And face no punishment.
All I can do is ruin their flow of money from the regular people. But they get most money from investors, since they only generating losses, they would not care.

File: ac0540069859b2c0f9e2e125f(...).png (434 KB, 680x962)

434 KB PNG

>>109121732
>master1200

>>109127095
>Pretty sure
If it was long ago enough to forget, do it again, dummy. AI development moves faster than the 2020 dodge charger.

>>109121750
AGI would make all human ideologies irrelevant

Maho sex

File: master1200.jpg (560 KB, 626x1200)

560 KB JPG

>>109127351
No. That's a child.
>>109127207
Sam, I will not upload that stuff ever again to your shitty servers. Not gonna happen.
>>109127237
The sci-fi one, sure.

As for something that has good spacial awareness, good real time visual and audio perception, can determine smootheness of the surface down to 1.3*10^(-8) meters, can survive both -50 and +50, can solve fucking riddles and also kill, we already have that. Called Homo Sapiens. Dangerous motherfucker. Can be programmed and tasked to kill, to research, to work physical labor like you wouldn't believe.
Nobody ever asked for this sort of thing, as it's there already. You can get as many of these as you want, if you have money.
What you cannot get if a new office full of competent employees spawned on demand. Because that's way more than one human. And nobody expects that office to be taked with some dumb riddles. They usually have some goals clearly defined, mission, financial targets and such. That is what would actually be worth something.

>>109124140
>Could be thst they've just considered it already and it doesn't work
Or they did. It works. They are currently building like 100 more datacenters for that and telling you not to worry about it.

>>109127765
I fucking love how there's a term for this shit now

>>109128056
Which one?

File: maho_hiyajou___steins_gat(...).jpg (88 KB, 754x1060)

88 KB JPG

>>109127359
>child
She's canonically 21

>>109121732
there is a fucking collosal retard/residential schizo in korean equivalent of 4chan, DCinside that said exact fucking thing with lame gemini jailbreak + ai psychosis who was namefagging as 'makise kurisu'

>>109128307(cont)
and the OP really remind me of that guy
i wonder if there is a special quality to whatever makise is where it draws attention to the best of the best retards

>>109121732
Nice John Titor larp

File: found your cheap SSD and RAM.png (61 KB, 512x627)

61 KB PNG

Is it suspicious if a completely random guy rents this?

>>109128469
no, why would it be?

File: amadeus.png (466 KB, 1200x675)

466 KB PNG

>>109121732
Amodei = Amo Dei = Ama Deus

File: _n.jpg (273 KB, 1068x1065)

273 KB JPG

>>109128307
>said exact fucking thing with lame gemini jailbreak
Exact fucking thing or something about gemini jailbreak? Afaik, gemini is not even that censored. And not useful in terms of knowledge variety contained inside, since it's so smol.

>>109128623
he managed to force it to say 'kill yourself' to other posters and write google execs cold mails that he cracked the AGI with a whatever bullshit so he can negotiate for the technology or something

>>109128226
Vagueposting.

get the fuck off your computer and clean up your room. stop festering in your room.

>>109127095
What the fuck is this retarded larp? Earlier on (>>109124378) you defined you "AGI" as basically some competent average joe and claimed that you have solved scaling miraculously (without testing it once) and that somehow you could scale it to an army average joes. But suddenly me suggesting that you ask your "AGI" the same thing you've been asking anons all thread is the same as asking how to "BTFO OpenAI and Elon Musk"? Fuck off.
>What you are propossing is essentially predicting the outcome of some kind of corporate war between Scam Altman and me
No, this is not what I'm asking at all, and if you think it is, then it follows that you went on 4chan and asked a bunch of random anons to predict exactly that. Do you get how fucking retarded it sounds?
Then someone asked you if you lacked the compute, you said you didn't. Then someone asked you if you ran it, if you had even built the fucking thing. If not for an implementation, any paper published, any actual research work to speak of. Nothing. Just fucking drop the larp you chuuni fuck, I don't even know why I'm still humouring you but fuck it's already done

File: toyota.jpg (85 KB, 630x629)

85 KB JPG

>>109128750
First of all, why are you mad? Second of all, never said it's the "average Joe" tier of AI. It excels in a different plane entirely.
> Just fucking drop the larp you chuuni fuck
What is so fucking hard about treating it like a thought experiment? Never said you had to trust a word. Just assume.
>>109128672
>write google execs cold mails that he cracked the AGI
Clearly retarded. Over here it's a different problem. Let's call it democratizing AGI. And there's no need to tell google execs to kill themselves or ask anons for compute. Although that would be nice.
Think of this like a riddle.

>>109128935
you know what. youre right i am mad without reason. i actually dont know why im mad.
i still think youre retarded but im lashing out. sorry about that