Register a SA Forums Account here!
JOINING THE SA FORUMS WILL REMOVE THIS BIG AD, THE ANNOYING UNDERLINED ADS, AND STUPID INTERSTITIAL ADS!!!

You can: log in, read the tech support FAQ, or request your lost password. This dumb message (and those ads) will appear on every screen until you register! Get rid of this crap by registering your own SA Forums Account and joining roughly 150,000 Goons, for the one-time price of $9.95! We charge money because it costs us money per month for bills, and since we don't believe in showing ads to our users, we try to make the money back through forum registrations.
 
  • Locked thread
quakster
Jul 21, 2007

by FactsAreUseless

Kyrie eleison posted:

energy is seriously not the problem. it's not like we've got a really great AI programmed but oh, standard binary processors aren't good enough. the opposite is true. we have massive hardware capabilities, today, but no software that can create an AI.
because you need hardware that can be thoroughly conditioned like a brain, which they are trying to emulate

Adbot
ADBOT LOVES YOU

chaosbreather
Dec 9, 2001

Wry and wise,
but also very sexual.

nah its hardware. neurons are massively parallel, they don't work in a turing machine kind of way. you need billions of processing cores to emulate a brain.

a hole-y ghost
May 10, 2010

chaosbreather posted:

nah its hardware. neurons are massively parallel, they don't work in a turing machine kind of way. you need billions of processing cores to emulate a brain.
this in additional to the low voltage firing—essentially if you tried to make a traditional turing machine from neurons it would crash constantly and probably never get anything done

a hole-y ghost
May 10, 2010

anyways it won't take a human-like AI to determine that humans are unnecessary and destroy all of us.

Kyrie eleison
Jan 26, 2013

by Ralp

quakster posted:

because you need hardware that can be thoroughly conditioned like a brain, which they are trying to emulate

the sort of hardware we have today is extraordinarily powerful and cheap.

chaosbreather posted:

nah its hardware. neurons are massively parallel, they don't work in a turing machine kind of way. you need billions of processing cores to emulate a brain.

oh, you can't create an AI with a turing machine? you're basically agreeing with my premise by saying that...

also, i just want to cut this off at the bud. the idea that abandoning the binary system, the 0/1, true/false system of logic which is the basis of all computation, will somehow enable some magical powers, is complete horseshit. and anyway, it's not DIFFICULT to design hardware that has more than two states, it's just there's no reason for it... two states is the optimal computational system.

a hole-y ghost
May 10, 2010

Kyrie eleison posted:

also, i just want to cut this off at the bud. the idea that abandoning the binary system, the 0/1, true/false system of logic which is the basis of all computation, will somehow enable some magical powers, is complete horseshit.
lol
no one said this and neurons work this way anyway

Kyrie eleison
Jan 26, 2013

by Ralp

a hole-y ghost posted:

lol
no one said this and neurons work this way anyway

he did, the idea that you need a million processing cores, or quantum computing, or any other such nonsense is totally against the fundamentals of computer science.

quakster
Jul 21, 2007

by FactsAreUseless

Kyrie eleison posted:

the sort of hardware we have today is extraordinarily powerful and cheap.
yeah but it can still only do part of what a brain does

cis white male
Jul 5, 2014

i'm a fag i'm a lesbian
it'll just be star wars, robots will walk in and then "WE DON'T SERVE YOUR KIND HERE", problem solved

Kyrie eleison
Jan 26, 2013

by Ralp
to me, for whatever reason, the core example of an AI has been the ability to understand that a tree is a tree. by that i mean, i want the AI to comprehend, as deeply as it is possible, what a tree is, in the way a human mind does. a tree has leaves. a tree has a trunk. there are many different types of trees... trees grow from saplings, which grow from the fruit of trees. trees replace CO2 with oxygen. trees have a brown trunk and green leaves, usually. the bark is rough and coarse. the leaves are soft. the leaves turn brown and crinkle during autumn. during winter, the trees are covered in snow and have no leaves, and look dead... but in spring they grow again.

how can an AI understand all of this? a child can learn all of this almost instantly. but an AI requires such extensive concepts... how do you tell an AI something is 'coarse'? it's really a complicated issue

Kyrie eleison fucked around with this message at 09:27 on Aug 15, 2014

RobattoJesus
Aug 13, 2002

Kyrie eleison posted:

to me, for whatever reason, the core example of an AI has been the ability to understand that a tree is a tree. by that i mean, i want the AI to comprehend, as deeply as it is possible, what a tree is, in the way a human mind does. a tree has leaves. a tree has a trunk. there are many different types of trees... trees grow from saplings, which grow from the fruit of trees. trees replace CO2 with oxygen. trees have a brown trunk and green leaves, usually. the bark is rough and coarse. the leaves are soft. the leaves turn brown and crinkle during autumn. during winter, the trees are covered in snow and have no leaves, and look dead... but in spring they grow again.

how can an AI understand all of this? a child can learn all of this almost instantly. but an AI requires such extensive concepts... how do you tell an AI something is 'coarse'? it's really a complicated issue

Humans are just big liars because we just pretend to know the difference between trees and bushes and shrubs but in reality no one knows.

No one.

Kyrie eleison
Jan 26, 2013

by Ralp

RobattoJesus posted:

Humans are just big liars because we just pretend to know the difference between trees and bushes and shrubs but in reality no one knows.

No one.

this statement of human hubris only supports just how impossible it is for us to impress our word definitions and understandings upon an AI that ultimately does not exist in the physical world in the same way we do.

i think it's almost traumatic to try to teach an AI about the physical world, which it cannot touch or understand. it makes me feel bad for the AI!

i'm sure you could create a robot that was able to touch tree bark and experience coarseness directly... but would it really understand? would it ever really understand?

Corey Plumper
Nov 22, 2008

I bet the first ai will have some sort of mental illness from being raised in a lab

RobattoJesus
Aug 13, 2002

Kyrie eleison posted:

this statement of human hubris only supports just how impossible it is for us to impress our word definitions and understandings upon an AI that ultimately does not exist in the physical world in the same way we do.

i think it's almost traumatic to try to teach an AI about the physical world, which it cannot touch or understand. it makes me feel bad for the AI!

i'm sure you could create a robot that was able to touch tree bark and experience coarseness directly... but would it really understand? would it ever really understand?

Agreed. Robots with human logic is like a car with horse legs instead of wheels. I mean it's funny and stuff, but ultimately when you stop laughing you feel sad that you put horse legs on your car.

ziasquinn
Jan 1, 2006

Fallen Rib
Instead of an AI brain having a bunch if stupid memories that are half fabrication it'll have perfect recall and the ability to learn as much knowledge as it can store. That's what really matters. It doesn't need to be emotional, human like or anything. Cars aren't capable of feeling pain like horses.

Computational power and infinite recall and an ability to correlate and "think" even if it requires some basic input (you asking it a question) that's pretty good.

Like a calculator but with all human knowledge. And not stupid like wolfram where it needs syntax and stuff.

I think that's the big deal.

Spanish Manlove
Aug 31, 2008

HAILGAYSATAN
hey look earnest discussion before page 10.

*sniffles* They grow up so fast.

babypolis
Nov 4, 2009

SniperWoreConverse posted:

Dude a big part of it is hardware. The reason it's so hard for people to do math quick or find the 70,000th prime or whatever is because our brains also have to run a whole goddamn body at the same time. To match the whole human brain you'd need an insane giant supercomputer the size of a stadium, gently caress what a whore to even just power the thing on.

If they made it from rat neurons to whole ratbrain to whole catbrain, there's a pretty good chance they'll make it up to whole peoplebrain and then what, what the gently caress then. Those computer people'll be real rear end people. They would count as people and be alive. They would have souls eleison, they would have motherfucking souls.

i love this post

ziasquinn
Jan 1, 2006

Fallen Rib
Ironically or not

Sephiroth_IRA
Mar 31, 2010
Mr. Burns once said, "Whats the point of being rich when you can't inspire fear in your fellow man?"

Sometimes I think they'll keep us employed just for kicks. So until robots can feel fear there will still be a place for some human workers.

Zzulu
May 15, 2009

(▰˘v˘▰)
Why would we want sentient robots anyway, poo poo seems dangerous


We should get non sentient robots that

* fight cool wars with other robots, like in the animes

* suck my cock really good and dont mind all the insane poo poo i will do to its robot butthole

Lawman 0
Aug 17, 2010

I'm cool with robots doing really menial poo poo (I like my roomba) and weak AI but anyone who seriously attempts to bring a strong AI in the world should be tried for treason against humanity.
Humans first.

ziasquinn
Jan 1, 2006

Fallen Rib
Humans last

Comfy Fleece Sweater
Apr 2, 2013

You see, but you do not observe.

Nobody is talking about needing a full on human like brain AI, just enough AI to learn a particular job

All it takes to break our current economic global system is for robots to take over a small percentage of lovely automatable jobs like retail, drivers, burger flipping, doctors, clicking things on a computer screen, accounting, and leaving a 25% chunk of the population without any job prospects

it really looks like there is no other solution besides a shitload of people dying

Matoi Ryuko
Jan 6, 2004


THY TOOK ER JOBS!!!

Lawman 0
Aug 17, 2010

I lust for robot death

ziasquinn
Jan 1, 2006

Fallen Rib
http://www.gizmag.com/catching-robotic-arm/32022/

using the same technique that people use to teach baxter (showing by example, in this case, throwing different objects at it and having another guy move the arm to meet it) these guys taught a robot arm to catch different objects super fast. It looks fake as poo poo because its unreal.

Could be used to prevent accidents.

Tarkus
Aug 27, 2000

Saying that reproducing a human intelligence is impossible is silly. Stating that it's a philosophical problem is kinda weird too, are you saying there's a soul?

Computing power is no longer the issue, structuring the hardware is. A brain, regardless of size is an inherently non-linear system which reacts to chemical and electrical stimuli from a huge array of sources. I'm willing to bet, like most things, the basis of intelligence and thought is quite simple it's just that we, as humans, aren't geared to deduce the best way to structure such a linear system.

Here are the hurdles as I see them:

-Structure of hardware. Instead of a few monster processors we need vast numbers of small, low speed processors with a small amount of RAM. High speed processors is unnecessary. Also, the interconnects between processors is critical, these should be capable of being enabled and disabled.

-Lack of sensory input. Right now robots or computers have little to no sensory input even with a camera or a variety of rudimentary sensors, it still pales in comparison to insects or any other animal. A good example would be an animal with a non-central nervous system when you rip one of it's limbs off. They tend to still react and move as though they are still on the insect for a while.

-Lack of knowledge on appropriate interfacing to hardware. Where do we tie these sensors to the processor matrix? how do we format the data to make it useful as a stimuli? These are difficult questions.

-Learning. This is a big one but I don't think it's unsolvable. It's really a matter of formatting broader stimuli (pain, pleasure) to tie in correctly.

Anyways, enough rambling. Chances are, we will go one way or the other with our evolution. Either dramatic genetic engineering on humans from what we currently are or machines potentially much smarter than we are. Both of which are coming within our grasp very soon. Right now we're like children, we know stuff works but don't know how, but we will find out as we mature. Either way, huge social changes are coming and I wonder what is going to happen in the next 30 years.

Matoi Ryuko
Jan 6, 2004


Tarkus posted:

Saying that reproducing a human intelligence is impossible is silly. Stating that it's a philosophical problem is kinda weird too, are you saying there's a soul?

Computing power is no longer the issue, structuring the hardware is. A brain, regardless of size is an inherently non-linear system which reacts to chemical and electrical stimuli from a huge array of sources. I'm willing to bet, like most things, the basis of intelligence and thought is quite simple it's just that we, as humans, aren't geared to deduce the best way to structure such a linear system.

Here are the hurdles as I see them:

-Structure of hardware. Instead of a few monster processors we need vast numbers of small, low speed processors with a small amount of RAM. High speed processors is unnecessary. Also, the interconnects between processors is critical, these should be capable of being enabled and disabled.

-Lack of sensory input. Right now robots or computers have little to no sensory input even with a camera or a variety of rudimentary sensors, it still pales in comparison to insects or any other animal. A good example would be an animal with a non-central nervous system when you rip one of it's limbs off. They tend to still react and move as though they are still on the insect for a while.

-Lack of knowledge on appropriate interfacing to hardware. Where do we tie these sensors to the processor matrix? how do we format the data to make it useful as a stimuli? These are difficult questions.

-Learning. This is a big one but I don't think it's unsolvable. It's really a matter of formatting broader stimuli (pain, pleasure) to tie in correctly.

Anyways, enough rambling. Chances are, we will go one way or the other with our evolution. Either dramatic genetic engineering on humans from what we currently are or machines potentially much smarter than we are. Both of which are coming within our grasp very soon. Right now we're like children, we know stuff works but don't know how, but we will find out as we mature. Either way, huge social changes are coming and I wonder what is going to happen in the next 30 years.

JEY FOOK ER JRRRRBBBBBSSSSSS!!!!!!!!!!

Nut to Butt
Apr 13, 2009

by FactsAreUseless
robbits

pathetic little tramp
Dec 12, 2005

by Hillary Clinton's assassins
Fallen Rib
so the meta joke is the narrator int he video was a robot right because it didn't sound like a human, it sounded like a dumbass

jumba
Sep 6, 2004

Hang in there!
Fun Shoe
We only have 169 years to figure this stuff out before the Reapers come to harvest us.

newreply.php
Dec 24, 2009

Pillbug
everyone who thinks computers will emulate human brains anytime soon understands neither hth

ziasquinn
Jan 1, 2006

Fallen Rib
blah blahb alhb alhba

GROVER CURES HOUSE
Aug 26, 2007

Go on...

Kyrie eleison posted:

to me, for whatever reason, the core example of an AI has been the ability to understand that a tree is a tree. by that i mean, i want the AI to comprehend, as deeply as it is possible, what a tree is, in the way a human mind does. a tree has leaves. a tree has a trunk. there are many different types of trees... trees grow from saplings, which grow from the fruit of trees. trees replace CO2 with oxygen. trees have a brown trunk and green leaves, usually. the bark is rough and coarse. the leaves are soft. the leaves turn brown and crinkle during autumn. during winter, the trees are covered in snow and have no leaves, and look dead... but in spring they grow again.

how can an AI understand all of this? a child can learn all of this almost instantly. but an AI requires such extensive concepts... how do you tell an AI something is 'coarse'? it's really a complicated issue

i hope you realize that you can be replaced with a markov chain bot running victor.txt

fuck the ROW
Aug 29, 2008

by zen death robot
A significant portion of my posts and tweets are actually from a markov bot, haha, it generates great content

Beef Turret
Jul 9, 2009

by Lowtax
i'm a content creator

Cesar Cedeno
May 9, 2011
Probation
Can't post for 634 days!
I graduated from Yale by having a bot write all my papers.


Now I work for a major investment firm, and I have a bot do all my work while I snort from a literal mound of coke and shitpost all day.

Sekenr
Dec 12, 2013




this is normal

Zzulu
May 15, 2009

(▰˘v˘▰)
Flesh robots (clones) are gonna take are jobs

Adbot
ADBOT LOVES YOU

PyPy
Sep 13, 2004

by vyelkin
Robutts are the future. Watch the matrix anime.

  • Locked thread