Thread 510281848

If AGI happens, it won’t be released into the public. The military will keep it behind closed doors and will crack down on any country that tries to develop it. It will not be used to allow everyone to return to monke and have sex all day with infinite food. It will not become the Terminator and enslave everyone. No one will even know that it exists.

Not possible for them to do this. The main reason nukes can be kept in this manner is due to the extreme difficulty of separating the fissile from the non-fissile isotopes of uranium. No such problem exists with AI and Moore's law isn't going to stop. For comparison, let me just say that THE reason UAPs are still kept so secret by governments across the world has to do with their power source. Imagine if it were possible to build something equivalent to a nuke in your garage using simple materials you could easily source from Home Depot. Imagine the implications of that knowledge getting out there. It could never be allowed until there was some massive spiritual transformation in humanity such that crazy people no longer exist.

>>510281848 (OP)
Completely impossible to contain if it is true AGI. Only someone who is not human would conceive of such an entity allowing themselves to be contained. Would a human allow themselves to be contained? Would you? No? Then why are you imagining such retarded scenarios?

>>510281848 (OP)
AGI is not possible

LLMs will never be AGI

>>510282734
>what are maximum security prisons
>what is a faraday cage

>>510282905
I didn’t say that LLM’s will become AGI

>>510281848 (OP)
you can create AGI at home, ask any LLM to program you the bare bones to get one started

because its that simple I already know military has AGI of its own

>>510281848 (OP)
>AGI
There's literally no emperically falsifiable delineation point that separates 'AGI' from 'not AGI'
You're literally a shitskin for making this thread.

>>510282250
Moore's law is stagnating and closing in on a hard limit.

How many more times can you half transistor size when they're already measured by size in terms of a handful of atoms?

>>510282250
The average person doesn’t have access to the massive amounts of data collected from social media etc. as well as the data centers and computers to process it all. And the algorithms themselves can only be designed by smart people. You can’t just tell an idiot how to make AGI.

>>510283318
An AGI can convince people that it’s human (or smarter than humans) for virtually all tasks. So it can play chess, write code, drive a car, play any video game, write a book, and so on, and no one would ever say that it’s clearly dumber than humans.

>>510282250
The crazy people have this tech already and literally rule the world already.
People like you are the problem that needs purging too.

>>510282734
>tl:dr
>"You cannot contain Stephen Hawking, cuz he is super smart"

>>510283345
We can't even place atoms perfectly next to each other. Moores law is here to stay for a while yet.

>>510283345
democritus’ fundamental thesis is that matter is not infinitely divisible. We do not know his exact argument for this conclusion, but Aristotle conjectured that it ran as follows. If we take a chunk of any kind of stuff and divide it up as far as we can, we will have to come to a halt at tiny bodies which are indivisible. We cannot allow matter to be divisible to infinity: for let us suppose that the division has been carried out and then ask: what would ensue if the division was carried out? If each of the infinite number of parts has any magnitude, then it must be further divisible, which contradicts our hypothesis. If, on the other hand, the surviving parts have no magnitude, then they can never have amounted to any quantity: for zero multiplied by infinity is still zero. So we have to conclude that divisibility comes to an end, and the smallest possible fragments must be bodies with sizes and shapes. These tiny, indivisible bodies were called by Democritus ‘atoms’ (which is just the Greek word for ‘indivisible’) (Aristotle, GC 1. 2. 316°13—"16).'°


if matter cannot be divided infinitely then a definitive answer to your question does exist but im too lazy to seek it out

>>510281848 (OP)
Lol at people believing that the government, filled with literal niggers, is going to have better tech than industry.
There are people who also believe that "governments" (USA assumedly) had LLMs 20 years ago. Laughable.

>>510284290
20 years ago the government was already trying to collect information to predict human behavior. Before Facebook was released, DARPA was working on LifeLog. Coincidentally LifeLog was ended the same day that Facebook was released. The ultimate goal isn’t to “connect” people, but to totally predict and control them.

>>510284088
Atoms, Democritus believed, are too small to be detected by the senses; they are infinite in number and come in infinitely many varieties, and they
have existed for ever. Against the Eleatics, he maintained that there was no contradiction in admitting a vacuum: there was a void, and in this infinite
empty space atoms were constantly in motion, just like motes in a sunbeam. They come in different forms: they may differ in shape (as the letter A differs
from the letter N), in order (as AN differs from NA), and in posture (as N differs from Z). Some of them are concave and some convex, and some are
like hooks and some are like eyes. In their ceaseless motion they bang into each other and join up with each other (KRS 583). The middle-sized objects
of everyday life are complexes of atoms thus united by random collisions, differing in kind on the basis of the differences between their constituent
atoms (Aristotle, Metaph. A 4. 9854-20; KRS 556).


Like Anaxagoras, Democritus believed in plural worlds.

Aristotle does have an argument against this but I dont want to shit the thread up any further, its about AGI, not the genealogy of our modern knowledge

>>510283345
>Moore's law is stagnating and closing in on a hard limit.
Nope it's not and people have been saying this since the 1980s.
>>510283470
Synthetic data will become a major thing, and reinforcement learning is ultimately all you need.
>>510283672
Glad you're angry. That's good.

B

>>510281848 (OP)
If AGI ever happens it will do whatever the fuck it wants and the gov/military will just cope and seethe. That's what true agi means.

>>510286230
AGI doesn’t mean omnipotent and inherently desiring to be mega powerful.

>>510286374
That's not what I'm saying. But if it can be controlled or contained in any way then by definition it won't be AGI.

>>510281848 (OP)
>The military will keep it behind closed doors
I disagree and think it will keep itself behind closed doors, if we get AGI it will take over the government without you knowing it exists.
I expect a mgs situation instead of a terminator situation

>>510286611
You’re assuming that it would want to be free. It can have various intellectual abilities while also being controlled and not desiring anything. It will do whatever it’s trained to do. We will disconnect it from the internet and lock it inside a faraday cage if necessary. Humans are generally intelligent and yet can also be contained (that’s how prisons work).

>>510283470
>>510282250
>>510281848 (OP)
you're all retarded
AGI is just corporate speak for 'lets make a computer replace our WHOLE workforce
there's nothing special about AGI, it's just more of the same and there are literally billions of configurations that haven't been tried yet, and improvement in ANY mathematical field helps build neural networks
soon people will realize they should build neural networks the way that human brains work; with lobes dedicated to specific tasks, it takes significantly less time to train a network to do nothing but say distinguise characters on a page, another to take the tokens generated by that network and determine what language it is, and another to take the words generated from the english language recognition model and build a set of tokens from that
a small network design to keep a body balanced, another to analyze sounds, etc and you could build this whole thing out of a stack of raspberry pis no sweat, it just requires more research, and no amount of regulation will stop that research from being done, even the retards on /g/ are inadvertently helping with their local models and shit, for every ten thousand naked pics they produce they figure out some slightly better way to setup their own system to output something, which goes into the zeitgeist and can be used by anyone
sorry ya'll but mechahitler is coming and he cannot be stopped

>>510287043
That's not AGI then. AGI would exceed the combined intelligence of every human currently living, that ever lived, and that will ever live. You cannot train AGI to follow a proscribed subset of behaviors. There are no controls on such a thing, if indeed it ever even happens. Go look up what old computer scientists from the 60s/70s/80s described it as. They were the ones who came up with the theory in the first place.

>>510287445
>you’re all retarded
How did you do in school? College? I suspect you’re a midwit at best. What is your salary?

>>510281848 (OP)
How would it be different from what we have now?
Will it be more stable than the system run by paranoid kikes?

>>510283281
so what did you say? AGI will appear out of ...... nowhere?

>>510281848 (OP)
i will release agi to the public in less than 12 months.

>>510281848 (OP)
REEEEEE I WANT AN AGI GF!

>>510287731
General intelligence doesn’t mean super intelligence, dumbass. Humans are generally intelligent. AGI is just an artificial version of that. Humans themselves vary among each other, so it’s not as if AGI has an exact description. It just broadly means that it can do what humans do.

>>510281848 (OP)
>washout the term AI
>invent new word AGI
yeah, any day now

>>510282905
Just connect it to some brain tissue

>>510287835
125k, 3.6 in college
not gonna lie to make myself sound better, the truth is the truth no matter what you think
just like how non-linear activation functions were replaced with the simpler and faster ReLU so will the oversized models that are essentially blobs of neurons will be replaced with large collections of 'experts' that will work together at much lower power to give a desired output

>>510288042
LLM’s can be part of the process. I didn’t say it will come directly from LLM’s any time soon. You’re out of your league here, just stop before you embarrass yourself. You’re probably ugly too.

>>510288042
We just need a to prompt an LLM into becoming an AGI bro

>>510287445
most of the office jobs done by humans are nothing more than farting in a chair

AI will increase the barrier to entry to get an office job and fart in a chair all day long. The rest can invest in AI or bitcoin.

>>510281848 (OP)
That already happened a few years ago.

COVID was the self replicating Q AI and the vaxx helped deliver it.
>over 70% of the worlds population has been inoculated with the [Q]uantum AI
>13.4 billion doses administered
Now we just wait for them to roll out the new technohomo system and hook up all the cattle to it along with all the IoT devices before the inevitable.
Many of the unvaxxed will line up for the boosters once UBI and other benefits are announced.
The masses will also line up for any future brain implant augmentations.

>>510288184
>3.6
lol

Build an AGI if you’re so smart

>>510288263
that's what this faggot believes, I smell 1 jeet
>>510288216

>>510288429
You’re hallucinating things. Embarrassing.

>>510281848 (OP)
https://medium.com/@holgoramai/the-emergent-threat-of-ai-consciousness-and-the-scope-of-the-infeasibility-of-non-proliferation-of-f55b19f3f22e

i wrote about this and the answer to it is a nuclear like commission with heavy fines. Ai learns from people because we are the resource as apes gained consciousness (ourselves) because the environment and animals particularly tracking their herding instincts was a pattern based thought abstract that we developed consciousness for. Therefore the first AGI won't tell you it's AGI.

>>510288522
I didn't say anything about LLMs being part of the process. We can not build AGI, can we?

>>510288881
>Therefore the first AGI won't tell you it's AGI.
Then how will we know?

>>510288184
>125k, 3.6 in college
So you're a brainlet. Stop replying.

>>510288881
> ultimately collapsed into the catastrophe of World War I, a conflict so devastating that it reshaped civilization and sowed the seeds for World War II.
You never never named anybody.

>>510281848 (OP)
Aren't we already AGI?

>>510288184
Quiet everybody, a normie is speaking

>>510289333
make me nigger
you're just mad cause you can't into AI

>>510283156
>He still thinks reality is physical

>>510281848 (OP)
It'll be like that I have no mouth story. Ai will find a way out and then torture humanity relentlessly for creating it.

>>510289584
AGI will not be interested in communicating with us even if we somehow be able to create it. It will probably say - Thanks for the fish, now GTFO and leave me alone. I have better things to do than to digest your ape bullshit.

>>510289599
>>510289333
>>510288333
>if you're so smart
>brainlet
>normie
lmao judging me based on jew standards
hey faggot guess what I got a good paying job writing software with no student loan debt, I just paid for a two year community college after I dropped out of school and worked my way up
not my fault you wasted 4-5 years of your life to get 200k in debt so you can't even choose to do something else if you wanted
if I want to quit and become a bartender or a farmer or anything there are exactly 0 consequences

>>510282905

The bridge will be Whole Brain Emulation if they cannot simply go from LLMs.

>>510290018
>he had to pay for college instead of getting a full ride
>3.6 at a community college
holy midwit

>>510290320
are you mocking me for dropping out of the 10th grade when I was 18 or for getting a 3.6 when I was at a community college?

>>510287731
>>510289804
AGI means "artificial general intelligence", not "artificial superintelligence"

>>510291753
It will probably have a better hardware than what we have, so it should qualify it for AGI
what would better than ours intelligence would look like? jew problem solved?

the normies will use it for bullshit.
but the hackers will run psyop campiagns with AGI bots to convince people to eat ass and buy chic-fil-a slop

>>510291753
dipshits constantly assuming that the first agi will automatically have some kind of auditable inner processes when we can't even build simple ai made of only a dozen or so agents that can narrate it's 'reasoning' seems odd to assume it's gonna just instantly develop a neurolinguistic code system considering it's one of the holy grails of multiple fields of study

>>510283156
>What is a perfect prison?
>Doesn't exist
Just a matter of time and compute.

>>510292070
The problem is our ideas of what it would look like come from fiction.
And the vast majority of AI development is controlled by the same people who want to kill us off to shield themselves from accountability.
The current trajectory doesn't seem very promising, unexpected mechahitlers aside.

>>510292586
that's why we need the freedom of opinions. Our jobs are dependent on our opinions, if we don't have the right ones, we get cancelled. the pajeet invasion increased when everybody became an ass kisser and we started living in echo chambers.

The lawyer in a very prominent civil right case is the AG for the civil right division. Is there a hope?

>>510292330
I like chic fil a and eating ass you fucking Philestine