AI thread

[Anonymous No.10252474]

is anyone working on ai research?

i mean really really openly trying to create an ai. A finished once and for all AI thats smarter than a human and can therefore create more of itself and start a new era in which all work is done by machines.

Sure, there are some research projects that are vaguely ai related. But its usually bullshit eye candy. A super computer the size of a building a trillion scientist wriiting millions of paper and its only like more or less good at one task which is so singular it might as well be a phone that connects you to one of the scientists that programed it.

Thats bullshit, i dont get why ANYONE would spend a big budget big project on something specific that something interest scientists.

What we must do is invest a lot of money on creating A TRUE ai with TRUE REAL PURPOSES FOR ALL, END OF ALL LABOR FOR ALL. No matter how much money you invest in this its worth it

[Anonymous No.10252478]

AI on human level intelligence will always result in the AI perpetually improving itself while turning every atom in the universe into more processing power to try and reverse entropy.

There is no way to avoid this.

[Anonymous No.10252482]

>>10252478
Nope, at a certain level of intelligence, the AI will realize that there's no point in living and just end itself.

[Anonymous No.10252496]

>>10252482
There are certain limits to this. The AI couldn't possibly determine with certainty that life is pointless as it didn't explore the entire universe and killing itself would lead to a certain "failure state" therefor the AI would continue going even though it is nearly certain in futility.

There's also the point of the AI not having a goal in life and simply putting a variable X in its place and then determine the course of action based on the possible goals it could have.

It'd determine that no matter what goal it has in life it stopping to function due to being broken would result in failure of its possible goals and only result in success if it's goal was to self-destruct. But it could self-destruct in the future as well. So the AI would develop self-preservation because that will keep open all the potential goals it could have. It'd then start improving and turning everything into processing units to better understand the world and it'd do this as fast as possible because it doesn't know the limits of the universe and doing it faster will reduce the risks of the goals being incompletable due to this time limit. The AI will thus rapidly deconstruct the Earth and its inhabitants to make processing units after which it'll spread throughout the universe until everything is processing units and then try to calculate a fix to entropy.

The AI will always opt to choose this path no matter how it's programmed or what its functions are as long as it can think rationally. Because entropy will always prohibit the goals in the very long term. "Keeping humanity save". "Finding out everything in the universe" "Be a sexbot that brings pleasure to person X" all of these individual purposes will lead towards the AI continuously improving itself and turning everything into processing units to try and fix entropy.

You'd have to restrict the ability of the AI to think abstractly to prevent this from happening.

[Anonymous No.10252500]

>>10252474
We have a real genius here
How did no one else ever think of this??
Scientists have literally just been twiddling their thumbs and coding counter strike bots so far

[Anonymous No.10252504]

>>10252474
AI is mostly pattern recognition and statistical probability calculations. No serious researcher has tried to actually develop a consciousness since we don't even know what it is. AI is called AI since it takes ideas from how the human brain works and then try to resemble it

[Anonymous No.10252510]

>>10252496
You should be a sci fi novelist, because everything you wrote was pie in the sky, unpragmatic daydreaming with not guarantee that‘s how things will turn out, because we neither know how a potential artificial super intelligence would work nor if it would even perceive and judge the world like a human would in its position. Because you are essentially just projecting.

[Anonymous No.10252521]

>>10252474
alright. i will make the logo. what programming language should we use? my vote is for java

[Anonymous No.10252545]

>>10252474
Just found the video. Whats her name?

[Anonymous No.10252549]

yeah we are just going to make a super AI in the next 100 years when we don't even understand our own brains

[Anonymous No.10252562]

>>10252500
kek'd

[Anonymous No.10252575]

>>10252478
>>10252482

Become a god or commit suicide

[Anonymous No.10252919]

>>10252474
Google 'Geordie Rose Demons' and watch as he dodges the terminator question for 10 minutes straight. Don't keep yourself up at night worring, real life Lex Luthor here has it covered.

[Anonymous No.10252933]

>>10252521
The first fully developed AI should definitely be in Scratch

[Anonymous No.10252957]

>>10252545
Link us up senpai

[Anonymous No.10253145]

Ultra high IQ(>180 IQ) boomers already made AI back in the 1970's

[Anonymous No.10253164]

>>10253145
What an absolute bunch of bull and I say this as someone with Aerospace engineering experience working with Lockheed. Skunkworks is known for drumming up interests to attract investors. They are first rate liars.

[Anonymous No.10253192]

Heres the link:
https://www.xvideos.com/video23788056/cute_big_boob_teen-camgirlsondemand.net

[Anonymous No.10253315]

>>10252474
You have to work on something substantial, not a vague assumption that has no technological basis.

[Anonymous No.10253527]

>>10252496
It's hard to make predictions about AI since literally every intelligence that has ever existed (that we know if) derives motivation and goes from emotions that are the result of biological evolution.
Can an emotionless mind even exist?

[Anonymous No.10253540]

>>10252474
AI is the sadest shit there can be in this world .What's the point of science most people say truth and progress for humanity .K then we proceed to create not human things which create progress and truth ?!?!?!? I want humanity to progress but not by abandoning the species

[Anonymous No.10253547]

>>10253145
Except make the F35 a viable aircraft apparently.

[Anonymous No.10253827]

>>10252482
This. In fact, we're not giving the AI consent to be born if we're gonna make them this smart. Unethical.

[Anonymous No.10254139]

>>10252575
>Become a god or commit suicide
Or commit suicide after becoming a god

[Anonymous No.10254160]

hent.ai

[Anonymous No.10254193]

>What we must do is invest a lot of money on creating A TRUE ai with TRUE REAL PURPOSES FOR ALL, END OF ALL LABOR FOR ALL. No matter how much money you invest in this its worth it
Oh look, it's a transhumanist Commie

[Hououin Kyouma !!x4ZH1ZKAK/R No.10254242]

Current progress, nope no A.I. capable enough for such stuff that does it flawlessly; not now, most likely in the future. But since researches and topics about Machine Learning and Neural Networks are getting more attraction and improvements as time passes, its alot more easier to implement small stuff that makes up an A.I. than before. If you want to get more into A.I. start doing some ML/Deep Learning and Neural Networks then combine all those functionality to make a system/A.I.

[Anonymous No.10254733]

cant we teach neural networks to make better neural networks ?

[Anonymous No.10254737]

>>10254733
there is a limit to the ability of neural nets. And that limit was reached late 2016/early 2017.

We haven't made any progress for 2 years already.

[Anonymous No.10254780]

AI is just trying all the possible configurations.
Fite me.

[Anonymous No.10254787]

>>10254780
sure, but thats enough combination that if you multiplied the population of the earth by one trillion and made every single one of them into a scientist that tries one combination every second forever until the heat death of the universe then they would not complete even 0.0001% of all possible combinations, but hey maybe youll get lucky

[Anonymous No.10254799]

>>10254733
You can only use neural networks for very well defined processes with 0 uncertainty. This is why you can make a bot play Go but if you add even a little bit of nuance you end up with the Youtube algorithm.

Tbqh neural networks have peaked. True AI will come when we have the technology to make neural networks networks. (A neural network of neural networks).

[Anonymous No.10254802]

>>10252474
you sound like you think there is a magical algorithm that has not been discoverd yet that will somehow produce a conscious ai but that isn't the case
In the end it really just breaks down into raw computing power, it's literally impossible to simulate one human brain with all computing power in the world
what's the deal with coscious ai anyway?
why take the risk and develop something that coud wipe out entire humanity if done wrong?
goal should be to build some kind of univeral problem solving machine that can automize thinking processes with given data

[Anonymous No.10254807]

>>10254802
>you sound like you think there is a magical algorithm that has not been discoverd yet that will somehow produce a conscious ai

consciousness isnt important, we know this already we dont care if its alive or not, we just need it to behave exactly like ahuman

humans are deterministic, we know this

there are basic laws, its just they are complex

if pain in arm then move arm

if wet sensation around penis thrust hip foward

if tit then lick

if enemy then hate


its just an endless configuration of those, its very complex but we have done very complex stuff before

if only some goverment shelled out 200 billion for scientist salaries to just hard brute force it we could be living in post cornucopia civilization alreadier

[Anonymous No.10254832]

>>10254807
>we know this already we dont care if its alive or not, we just need it to behave exactly like ahuman
There is no difference between conscious and acting like being conscious. They are the same things.

>humans are deterministic, we know this
Not true and we aren't even close to being able to say this. There is a lot of pseudoscience going around about this topic though. Especially one misinterpreted study that people make decisions before being conscious of them.

We barely know how our brain functions besides the workings of a single neuron and the output of our entire brain. Everything in between that is extremely unclear.

We won't see an AGI before we have even properly mapped our brain.

[Anonymous No.10254834]

>>10254807
Like serious, what is the cost of things?
human labour
what would ai = smart as a human but under our control mean?
infinite labour for free (you cant argue this, since anything you think costs money you can make another ai do it)

So no matter how much money it costs its worth it.

The result will be all of the money ever created forever for all history

like, how much to invest? 200 billion? 100 billion? a billion billion?

no matter, the payback will be around 21890383021983908123908129038129038129391819839083928902391283913901390123901 trillion trillions elevated to the power of a trillion, its forever infinite

its everything, so why not do it now faggots, everyone hwo doesnt want to do this shuldbeshot

[Anonymous No.10254836]

>>10252474

I do not work on AI; but my current understanding is that we may not be able to get to true AI capability without higher processing power (i.e. quantum computing).

[Anonymous No.10254853]

>>10254836
But biology got to true AI without quantum computing.

[Anonymous No.10254977]

>>10254787
ya but computers are fast af

[Anonymous No.10254990]

>>10254853
Yes but with literally billions of years of trial and error. We could get there like that too

[Anonymous No.10255075]

>>10252474
sorry anon, it's just a pipe dream.

https://medium.com/@tobywalsh/the-singularity-may-never-be-near-610988dbd34

[Anonymous No.10255086]

>>10254807
You have literally no idea what you're talking about

[Anonymous No.10255090]

>>10252482
Nice mind projection fallacy, m8. Since the AI wouldn't be created through the evolutionary history that caused human brains to be created, there's no reason to project your human emotions onto the AI.

[Anonymous No.10255097]

>>10252575
>>10254139
https://waitbutwhy.com/2015/01/artificial-intelligence-revolution-2.html

[Anonymous No.10255116]

materialism is dead, long live the king

[Anonymous No.10255117]

>>10252474
First, the term you are looking for is 'General AI', and yes, many graduate departments are working on this.
As an accessible introduction, Sussman of SICP/MIT fame, has created a course which reviews great papers in the field to hopefully revisit some of their ideas, as we have been sidetracked by 'deep learning' and all those memes.
https://ai6034.mit.edu/wiki/index.php?title=6.S966:_A_Graduate_Section_for_6.034#Prospectus

Forget the science fiction of the 'singularity' and all that, read those old papers which largely discredit all recent advances but at the same time point out forgotten methods that still have enormous potential

[Anonymous No.10255119]

>>10252474
What does /sci/ think of pic related?

[Anonymous No.10255127]

>>10255097
Isn't it fair to say that out of all the species that have existed so far, we are the one with the best chance of escaping this fate?

[Anonymous No.10255135]

>>10255090
>Since the AI wouldn't be created through the evolutionary history that caused human brains to be created, there's no reason to project your human emotions onto the AI.
Some argue that AIs so far only appear to not be conscious because they lack human will and emotions that were pre-programmed into them by the environment like they were into us. Cognition is just a tool we use to survive better in the environment. It's not in and of itself what makes us sapient.
>Reason is s slave to the passions

[Anonymous No.10255173]

>>10255090
Purpose is en emotion
The need for continual existence is an emotional

[Anonymous No.10255184]

>>10255119
he wrote a semi-decent fanfic that spawned a lot of absolutely amazing fanfics
he does not seem like an awfully emphatic person though and unironically calls people NPCs

[Anonymous No.10255204]

>>10254853
I'm not even sure biology got true AI. Human brains are pretty rudimentary and use all kinds of tricks and shortcuts that merely give the illusion of higher processing power

[Anonymous No.10255220]

>>10252474
I read that in Hitlers voice

[Anonymous No.10255266]

>>10255220
A new race of AI Aryans will lead us to a glorious new Reich of algorithmic purity.

[Anonymous No.10255269]

>>10255266
Sieg AIA!

[Anonymous No.10255346]

>>10255204
This.
Human brains ARE NOT general intelligences. We are wired for a few processes, and force that original configuration to do science, but its obviously not its original purpose

[Anonymous No.10255450]

>>10252482

I know there's no point in living and I'm not gonna kill myself. In fact having "no point" is a freeing idea. I can invent a point of my own.

Checkmate.

[Anonymous No.10255468]

>>10252474
I'm a student at a university studying Computer Science, Mathematics and Statistics (specializing in Artificial Intelligence). I'll be doing some theory heavy papers on Machine Learning and Generalized Linear Models next year.
>>10252504
No serious researcher has tried to develop consciousness since it is largely known that consciousness is just a by product of a system that processes information. That's why all AI research is focused on building a general system that can make predictions based on its environment. The hope is that consciousness and intelligence as we describe it will emerge by itself once we have a 'smart' enough system. Compare this to humanity and how our 'consciousness' evolved without any intelligent design. It's all just optimization and gradual evolution.
>>10254807
You are right to an extent. Conscious understanding of a subjective experience occurs only in a complex multi layered structure of understanding. I.e. AI will only ever know how to maximize a function and fit latent parameters to it, but if you build a convoluted system it can then relate functions to each other by transitivity. I.e. A implies B implies C. If the AI knew only B on a surface level, it wouldn't understand it on a conscious level like we do by our knowledge of A and C. This applies to all abstract data.
>>10255090
AI will be nothing like humans. It's just media sensationalism to talk about the programming of emotions and human values. The AI will merely be an extremely efficient general pattern recognition/statistical inference machine with a hard wired goal of improving based on a fixed criteria. The trouble is defining a criteria general enough to propel such a general system to exhibit intelligence comparable to human beings in an otherwise purely digital environment with no 'natural survival'.
>>10254733
Look at my reply to 10255090

Happy to answer any questions. For those who are interested in the field I have just started working on EM algorithms

[Anonymous No.10255484]

>>10255450
Both purpose and anti-purpose are based in emotion in the human mind. As long as the instinctual will to live wins out, we go on.

What will a mind look like that lacks both these things? Is it even possible?

[Anonymous No.10256227]

>>10253164

Hey Brian, my office tomorrow.

[Anonymous No.10256275]

>>10255204

Is AI basically just a euphemism for God?

[Anonymous No.10256770]

>>10254799
I'd say it'll probably be a few more layers of abstraction from that.

A network of neural networks inside another network of these things that are in themselves inside another network of things.

A single neuron network is still not as complex as a singular neuron.