[ 3 / biz / cgl / ck / diy / fa / ic / jp / lit / sci / vr / vt ] [ index / top / reports ] [ become a patron ] [ status ]
2023-11: Warosu is now out of extended maintenance.

/sci/ - Science & Math

View post   
File: 141 KB, 1020x870, Computer_Power_Per_1000_Dollars.jpg [View same] [iqdb] [saucenao] [google]
1087831 No.1087831 [Reply] [Original]

What do you think about this?

>> No.1087840

there's a nigger on the picture

>> No.1087856

I think it's gonna be awesome meeting Adam Selene.

>> No.1087863

>Exponential Growth
>Logarithmic Plot

>> No.1087870

That graph needs to be adjusted to account for the fact that when people use computers, they naturally limit their intelligence because they don't need to use it.

>> No.1087876

How the hell do you measure "calculations per second" in something like a neural network?

>> No.1087887

>>1087863

Yeah, how does that work?

>> No.1087895

>>1087863
That's still possible.

>> No.1087897

>>1087876
yeah exactly. pic is BS

>> No.1087908

youre telling me that current computers have the computational speed of a rodent?

>> No.1087914

This graph doesn't make it past 2000. Probably because in the last few years, we have plateau'd at around 3.5ghz

>> No.1087946

Kurzweil can suck my hairy balls

>> No.1087948

>>1087876
If I remember correctly (could be wrong), this is based on what we know about the computing power of the human retina. We have a pretty good handle on the computing power of a human retina, because we've built artificial retinas (though they are not yet implantable). So what people who make graphs like these do is assume that the computing power of the cortex per unit area is similar to that of the retina. Whether that assumption is valid or not I couldn't begin to guess.

>> No.1087956

>>1087914
>Implying CPU frequency increase is the only way to increase computational power

>> No.1087970

I sti have a receipt somewhere around here for 2mb of ram. 1400 dollars.

>> No.1087979

>>1087831
That curve fit is absurd. Whoever made that graph should be shot.

>> No.1087989

>>1087979
Assuming you are correct, murder is not a good answer to absurdity in curve-fitting.

>> No.1087993

Well Moore's Law has been accurate so far, and I want fembots, so GO KURZWEIL!

>> No.1087994

>>1087948
But the structure of neural networks are radically different from computers. They don't process information in a comparable way. Even computer simulations of a retina make huge simplifications and assumptions about it's structure. We can't even properly model the principal component of a neural network (the neuron) without making huge simplifications. Not to mention that neurons differ tremendously throughout different types of brain tissue.

>> No.1088019

>>1087994

Not only that, but there's a lot more to a brain than raw computing power. No computer chip can make sense of the data it processes and we don't even have the first clue of how to make a computer capable of understanding.

>> No.1088024

>>1087994
Well, yeah -- that's one of the big questions: how much abstraction can you get away with and still have a simulation that is good enough? But the fact that neurons work radically differently from computers isn't itself a show-stopper. Computing is general purpose. A sufficiently powerful computer ought to be able to simulate any analogue process.

>> No.1088034

>>1087970

yo mod... 2 mb of ram for 1400 dollars?

you can go to walmart and buy 2 gb for like... 20 - 40 bucks

>> No.1088035

>>1087989
1) Being shot often not fatal.
2) Murder is a legitamate response to someone spending all that time making that graph but not knowing the first thing about curve fits.

>> No.1088036

>>1088019
Ah, the old Chinese Room argument. I never could figure out the problem that is supposedly posits so I'll leave that one untouched.

>> No.1088046

>>1088024
>A sufficiently powerful computer ought to be able to simulate any analogue process.
Sure, but we just don't have those yet, so the graph in the OP is speculation at best

>> No.1088049

>>1088036
I feel like the person who made the Chinese room argument was an idiot. It is just so flawed. It's like he never took a step back and actually thought of how people think, about how they assign meaning to words.

>> No.1088051

>>1088019
If you can simulate a complete brain on a computer, we have an intelligent computer.

>> No.1088059

>>1088046
That I'll agree with. It wasn't entirely pulled out of someone's butt, but there is a fair bit of speculation involved.

>> No.1088062

>>1087887
>>1087863
Anyone who didn't immediately recognize this should get the fuck out of /sci/.

Also, anyone who uses a log plot for anything but a linear fit a gigantic faggot, and likely retarded.

>> No.1088065

>>1088051
To be fair, we may never have the processing power capable of simulating every atom in the human brain. We will need a simplification of the neurons and a good deal more information about how the human brain works. I predict we will learn a great deal about ourselves as we pioneer AI.

>> No.1088071

>>1088051

but thats the problem... the human mind is not something you can just simulate with our limited technology...

there are millions upon millions of things that a computer still cannot do that we can.

and there are still things about the brain we do not understand

... plus... am i the only one who believes your brain is a separate being from you... like you use it, but when you sleep it just kind of does its own thing, just to entertain itself... not you

>> No.1088079

>>1088071
The moment you said "limited" is the moment I stopped reading your post.

>> No.1088080

>>1088071
You are your brain. That is what I believe. (also, I'm a neuroscience major)

>> No.1088084

>>1088079

why is that so?

we do not have unlimited technology... we are still developing things every day... new ways of doing things... new tools we use

>> No.1088090

technology is things that make living, and doing things easier...

we are creating more and more things to complete that defenition

>> No.1088105

>>1088079

and im not saying we can't do it eventually..

just atm we do not have the technology to do so

>> No.1088114

>>1088084
It just made me think you are underage. You use of the word "limited" seemed to imply there is some inherent limitation of technology that you then completely fail to explain.

wtv

>> No.1088115

>>1088065
> we may never have the processing power capable of simulating every atom in the human brain.

We needn't. We can optimise all that shit out and simulate a neuron on a single laptop (today-laptop) EASILY.

>good deal more information about how the human brain works.

Not really. We'll just do it as the brain works. We can get the finer details retrospectively.

>I predict we will learn a great deal about ourselves as we pioneer AI.

Thanks, Einstein.

>> No.1088128

>>1088115
>We can optimise all that shit out and simulate a neuron on a single laptop (today-laptop) EASILY.
Not without making huge simplifications about neural function we can't. Today, anyway.

>> No.1088129

>>1087831
> What do you think about this?
I think when we're anywhere near to "computing power of a brain" we'll find brains are a lot more complicated than Kurzweil suspects.

>> No.1088135
File: 38 KB, 523x478, Reading.jpg [View same] [iqdb] [saucenao] [google]
1088135

>>1088071 ... plus... am i the only one who believes your brain is a separate being from you... like you use it, but when you sleep it just kind of does its own thing, just to entertain itself... not you

>> No.1088139

>>1088115
You see, there is this thing about good posts... reading them bit by bit misses the big picture. You taking my posts section by section kind of made you miss what I was saying as is apparent by your last comment.

also sage

>> No.1088147

software power doesnt follow computing power. What Intel Giveth, Microsoft Taketh Away.

>> No.1088148

>>1088114

i'm sorry if my choice of words may have been bad, but my point is still clear

we do not possess the technology to create everything that we imagine, but we do have the technology and the initiative to create better technology to create more than we can imagine...

>> No.1088151
File: 35 KB, 400x527, 1260430723416.jpg [View same] [iqdb] [saucenao] [google]
1088151

>>1088139
>compliments his own post

you gigantic tripcoding faggot, get out

also your post wasn't great. 3 sentences that show no great insight.

>> No.1088152

>>1088129
That is why we are going to need to simplify it. I wonder what the ethical issues will be of creating self aware programs and then deleting them because they are not what you were looking for.

>> No.1088157

>>1088135

idk... something i just think about

>> No.1088163
File: 73 KB, 1068x600, face11.jpg [View same] [iqdb] [saucenao] [google]
1088163

>>1088151
>looses trip to insult a trip
>implying you aren't samefag

>> No.1088165

>>1088157
That is rather ancient Egyptian of you isn't it?

>> No.1088170

>>1088128
I suppose the burden of proof is upon me, as the maker of the claim.

Guardian (quoting Henry Markram, who is not estimating, but is reporting on lab progress): To simulate a single neuron takes the computing power equivalent of a laptop.
http://www.guardian.co.uk/media/pda+media-events-conferences
Also see the TED talk.

Seed: a single simulated neuron is really the sum of 400 independent simulations.
http://seedmagazine.com/content/article/out_of_the_blue/P1/

>> No.1088178

>>1088152
This bothers me a lot, actually. The emphasis on human interaction virtually guarantees that if we do create artificial intelligence, we will have scrapped it several times over simply because we wouldn't recognize machine intelligence.

So much of our perceptions are tied to our physical systems, and our empathy toward creatures similar to us (predator mammals, then prey mammals, then maybe birds and reptiles, then fish, etc). Of course a machine creature will have none of these characteristics. If it were sentient, it would be very difficult to tell, unless it were *very* intelligent.

>> No.1088181

>>1088163
I don't know whether you're implying that I posted that post with the man and dog but I was busy finding my sources and I don't have any negative feelings to you. (Haven't read the post that self-compliments yet though, maybe I'll be pissed too soon :P)

>> No.1088182

>>1087831
That is a lot of extrapolation. It could plateau off as far as we know.

>> No.1088184

>>1088165

haha, i wouldn't know...

history not my best subject... its good to know i share a same idea of some of the greatest engineers of the past though..

though i couldn't build a teepee to save my life

>> No.1088186

>>1087831
>computer power per 1000 $
Even if you gave a mouse fuckings million I don't think he would do even a single fuckings calculation you shithead.

>> No.1088193

>>1088170
Yes yes yes I know all this. The point is that every singe simulation to date conveniently assumes constant axonal and dendritical circumference which raises questions about the ecological validity of these models. These assumptions mean that the weighing of input from dendrites with different lengths is radically different form actual neurons. Today's simulations are approximations. (I love how that sounds)

>> No.1088194

>>1088139
If the claims within the sentences of your posts can't be judged on their own merit then you're obviously not writing for factual accuracy. You should be able to defend every claim you make on its own, so long as the original meaning is preserved (if I have altered the meaning then point it out). This isn't poetry.

>> No.1088199

>>1088186

i lol'd so hard at this.

>> No.1088206

>The point is that every singe simulation to date conveniently assumes constant axonal and dendritical circumference

That doesnt sound right. Or true.

>> No.1088207
File: 26 KB, 512x384, 1271624440737.jpg [View same] [iqdb] [saucenao] [google]
1088207

>>1088193
Hm, I wish you had been more specific about your argument to save me finding some sources. I don't know enough to address your claims, but I'll be sure to look into that. Thanks.

Is it likely that the assumptions you named will have a large impact on neuron function?

>> No.1088213

>>1088206
Show me a simulations that factors in variability of axinal and dendritical width.

>> No.1088225

>>1088213
a simulation*

>> No.1088232
File: 157 KB, 600x450, expo comp 4.png [View same] [iqdb] [saucenao] [google]
1088232

>>1087831
That graph is calculations per second for $1,000.

Calculations per second, in supercomputers, reached top-estimates of human level in 2009.

All that's missing for an AI as intelligent as humans, is the software.

>> No.1088243

>>1088232
I disagree. Computers do not work the same way as brains. Simulations require A LOT more computing power than the thing itself takes (my computer can barely emulate a PS2 despite having better hardware than a PS3), and if you plan to make intelligence native, you have to completely redesign the computer's architecture.

>> No.1088249

>>1088184
The ancient Egyptians scraped out the human brain with a hot poker through the nose during the mummification process. The other organs were taken out and put in jars to be stored in the tomb. The liquid remamnents of the brain however were discarded because they didn't know it what the brain did. They thought it was useless.

>> No.1088251

>>1088207
Okay here's a list of things commenly left out neuronal simulations:

>variability of axonal and dendritical width
-Variability of membrane permeability across the neuron
-The entire fucking metabolism of the neuron. Most simulations just assume it runs by itself
-Variability in receptor types across synapses (which also effects the weight of input from different dendritical connections with other neurons)

>> No.1088265

>>1088193
>ecological
???
Either that was a typo or I know nothing about neurology.

>> No.1088277

>>1088265
I was thinking about asking the meaning of that, too :P

>> No.1088288

>>1088265
>neurology
facepalm.jpg

neurology is to neuroscience as engineering is to physics

But just to be clear: The ecological validity of a simulations says something about the relevance the simulation has outside of a theoretical framework. It's a measure of how "real" the simulation simulates what it's supposed to simulate.

>> No.1088292

>>1088232

this is utter nonsense.

the brain has between 10 and 100 trillion neurons.

that's 10-100 trillion microprocessors working in parallel.

supercomputers are still dealing in the thousands.....possibly millions. were many orders of magnitude away from even coming CLOSE to the processing power available in the brain.

>> No.1088296

>>1088288
replace "real" with accurately*

>> No.1088318

>>1088034
This was 1986, though I could be off by a few years. Hard to remember every purchase. The only reason I still have the receipt is because the 1mb board it replaced got put into the box it came in, and the receipt went in with it.

>> No.1088325
File: 150 KB, 346x520, emmanuelle2.jpg [View same] [iqdb] [saucenao] [google]
1088325

supercomputers are already smarter than cats

http://arstechnica.com/science/news/2009/11/ibm-makes-supercomputer-significantly-smarter-than-cat.a
rs

>> No.1088334
File: 207 KB, 600x450, expo comp 6.png [View same] [iqdb] [saucenao] [google]
1088334

>>1088243
>>1088292

Human brain does around 10^16 calculations per second.

http://www.foresight.org/nanodot/?p=3051
http://www.merkle.com/brainLimits.html
http://movementarian.com/2006/08/18/flops-mips-watts-and-the-human-brain/
http://www.transhumanist.com/volume1/moravec.htm

Current supercomputers do 10^16 calculations per second.

we do not need to 'simulate every neuron. or something like that. brain does a certain # of operations every second. supercomputers can do that same number.

all that's missing is the software.

>> No.1088349

>>1088334
Why did you cite me? You didn't reply to anything I said, you just repeated the position that I was opposing with no new evidence or argument at all.

>> No.1088357

>>1088334
>Human brain does around 10^16 calculations per second.
Goddammit. You didn't read any posts in this thread did you

>> No.1088362

>>1088349
Yes he did, you are just too dumb to understand it

>> No.1088371
File: 14 KB, 380x360, implied facepalm cat.jpg [View same] [iqdb] [saucenao] [google]
1088371

>>1088325
and da vinci invented the aircraft in 1480

>> No.1088387
File: 30 KB, 274x418, At1st.jpg [View same] [iqdb] [saucenao] [google]
1088387

>> No.1088393

>>1088292
lol, thousands. try millions of transistors, and video cards go into the billions. currently there are 6-core single die HOME CPU's. the first graph of this post is also wrong, its to slow, the growth is actually alot faster then that. im betting by next year this time we will start seeing 8 or 12 core single die CPU's.
so with current trends in CPU power, we "should" reach the processing power of the human brain in 10-15 years, tops.

>> No.1088400

>>1088371

don't you trust IBM?

>the group's massively parallel cortical simulator, C2, now has the ability to simulate a brain with about 4.5 percent the cerebral cortex capacity of a human brain, and significantly more brain capacity than a cat.

>The simulator, which runs on the Dawn Blue Gene /P supercomputer with 147,456 CPUs and 144TB of main memory, simulates the activity of 1.617 billion neurons connected in a network of 8.87 trillion synapses

but I think It needs more energy than a cat brain LOL

>> No.1088405

>>1088393
> I am wildly optimistic about our understanding of the brain

>> No.1088409

>>1088393
Goddammit. You can't quantify "processing power" the same way for a human brain as for a CPU in a computer.

>> No.1088448

>>1088400
but scientists dont even know how neurons work yet,
in the current form, this "brain" is just like a wooden aircraft built by one of those cargo cult tribes in the pacific.
http://en.wikipedia.org/wiki/Axon_guidance

>> No.1088456

>>1088409
>>1088405
then this entire thread is invalid, but wait and see.

>> No.1088490

>>1088448
>but scientists dont even know how neurons work yet,
That's not true. We have a pretty solid understanding of individual neural function. It just gets complicated when you put lots and lots together.
>in the current form, this "brain" is just like a wooden aircraft built by one of those cargo cult tribes in the pacific.
wat
>http://en.wikipedia.org/wiki/Axon_guidance
Again, wat. I don't see how this is related.

>> No.1088515

I think the most efficient way to recreate the human brain would be to invent a whole new type of circuits (some kind of analog circuits that can modify their own connections, maybe).
Also, needs more understanding of the human brain.

>> No.1088530

Please everyone who is not a neuroscientist or a mathematician working on the brain field, SHUT THE FUCK UP.

kthxbye

>> No.1088539

>>1088490
>http://en.wikipedia.org/wiki/Axon_guidance
>Again, wat. I don't see how this is related.
well, it kinda has to do with LEARNING AND BEING SMART N SHIT!

>> No.1088541
File: 36 KB, 640x347, expo mem.jpg [View same] [iqdb] [saucenao] [google]
1088541

>>1088349
>I disagree. Computers do not work the same way as brains.
This doesn't matter. Parallel processing and whatnot, like a brain has, is tons more efficient. But it doesn't matter, because with our current style of computers, we've still caught up in speed. I mean a human brain runs on like 20 watts but can do an insane 10^16 operations per second. That's thanks no doubt in part to the parallel structure. Still doesn't matter. Current comps are just as fast! less efficient, consume a ton more power, yea, but we've caught up in raw speed.

>>1088409
>You can't quantify "processing power" the same way for a human brain as for a CPU in a computer.
Yea, you can. Operations per second means operations per second. An inch is an inch. Operations per second is a constant measure.

A computer does tons of operations per second one way. A brain does it another way. But they're both measurable in how many operations per second they can do.

Its like a human running a race vs a bicycle riding the same distance. Both travel a certain speed. One's faster than the other. Both travel the distance in different ways. But they both have a certain speed. The speed is measurable and comparable. Don't be silly.

>> No.1088551

>>1088530
But not really though. It has more to do with fetal brain development.

>> No.1088556

>>1088530
Read a fucking journal faggot, not 4chan. KTHXBAIIIII

>> No.1088558

>>1088515
http://en.wikipedia.org/wiki/Memristor here you go, basically a transistor that can store more then a 1 or 0.

>> No.1088560

>>1088541
Define "operation" in terms of neurotransmitters and membrane potentials.

>> No.1088564

>>1088551
no
>Neural development comprises the processes that generate, shape, and reshape the nervous system, from the earliest stages of embryogenesis to the final years of life.

>> No.1088567

>>1088362
Go on...

>>1088541
You only took the first half of my argument. That sentence wasn't meant to be an argument on its own.

>> No.1088574

>>1088560
I don't know, read and/or ask the guys I got the source from.

http://www.foresight.org/nanodot/?p=3051
http://www.merkle.com/brainLimits.html
http://movementarian.com/2006/08/18/flops-mips-watts-and-the-human-brain/
http://www.transhumanist.com/volume1/moravec.htm
there's more human brain OPS estimations on google as well. 10^17 is the very high-end estimate.

>> No.1088587

>>1087831
Why are "All mouse brains" not on the vertical scale, but "All human brains" are?! This is an outrage!
You are clearly discrimination against mice, you racist.

>> No.1088592

>>1088564
Yes. Axon guidance is only a subfield of neural development and plays a major role during fetal development. There isn't any major axonal restructuring postnatally.

>> No.1088595

except that a brain isnt a computer, and we're not that fast at math as a regular computer.

How the hell you would program a pineal gland, or a pre-frontal lobe into a chip?

>> No.1088601

>>1088567
You have no fucking argument. You're just saying 'i have an opinion and that opinion is you have to do this this and this before an intelligent AI is possible.'

There's no evidence to suggest
>and if you plan to make intelligence native, you have to completely redesign the computer's architecture.

IF I wrote the right software, back in 1980, we could have a human level AI back then. Bu it would be incredibly slow, and would take decades to 'think'.

All we need is the raw speed, and the right software. We have the raw speed now.

>> No.1088615

>>1088587
We don't know how many there are because they like to stay anonymouse.

>> No.1088617

Updated version please?

>> No.1088634

>>1088601
> human-level intelligence
Yeah... ok bro. I'd like to see how intelligent we consider you after we cut off your arms and legs, pluck out your eyes, puncture your eardrums, pinch off your nose, and remove your tongue.

ITT: intelligence is just measured automatically by Kurzweil-worshipping faggots

>> No.1088643

>>1088592
according to your model 30 year olds run around with the brain and intelligence of 5 year olds. you really want to continue with this bullshit?

>> No.1088644

>>1088595
its an organic analog processor. everything started off tiny and with a very simple structure, so we should "eventually" be able to reconstruct the process. its not like complex life just blinked into existence and works by magic.

>> No.1088659

>>1088574
>the guys I got the source from
I didn't look at them all but from what I read these are wild claims based on models of simplified neural function from commercial corporations.

If you have actual journal articles I'd be happy to take a look at them.

>> No.1088678

>>1088643
... you just don't quit do you.
>according to your model 30 year olds run around with the brain and intelligence of 5 year olds
Learn the difference between <span class="math">synaptic[/spoiler] modifications and <span class="math">axonal[/spoiler] modifications.
>you really want to continue with this bullshit?
I think I know a bit more about this stuff than you do so yeah... I do if you push it. I'm a neuroscience major.

>> No.1088711

so if we did create a self aware AI, what would we do with it? ask it questions? put it in a giant room with a bunch of puzzles and give it extensions to manipulate its environment?

>> No.1088721

>>1088711
sex robot

>> No.1088727

>>1088678
nobody gives a shit who you are. Axon guidance still happens in adult brains. this is fact.

and "our computer is smarter than a cat" is a smug overstatement by computer scientists who want to get publicity.

>> No.1088732

how would we test it? would an AI computer experience time as going really fast or slow?

maybe we could combine it with our biology somehow, like take a brain from some hobo and hook it up to a machine or maybe grow a few more layers of neocortex in the lab and hook it up to the internet

>> No.1088763

>>1088727
>Axon guidance still happens in adult brains. this is fact.
No fucking shit sherlock. The point is that it doesn't play a major role anymore and the article was completely unrelated to your post.

>and "our computer is smarter than a cat" is a smug overstatement by computer scientists who want to get publicity
at least we agree on something

>> No.1088803

i dont know if we would ever get to simulate the brain on a computer from just hardware and circuits, maybe the interconnectedness and web of the neurons but the parts of the brain are a product of evolution and have specific functions, an AI with completely nonbiological systems would be something completely foreign

>> No.1088885

>>1088601
Why do you start by saying I have no argument then continue by replying to my argument? I would love to understand how you've managed to pull that off.

>There's no evidence to suggest
>>and if you plan to make intelligence native, you have to completely redesign the computer's architecture.

It's common sense. You obviously did not understand the significance of the PS2 emulation example, so let me try to explain again. The computing power required to drop an object is zero. You just drop it. The computing power required to simulate dropping an object is non-zero. First you must simulate the relevant parts of our universe, then you must simulate objects to be dropped, then, in the simulated universe, you can drop the object and see how the simulated universe responds. If you were to simulate a computer simulating objects being dropped in our universe, you'd require even more computational power than the original task, because you have to simulate all of the parts of the computer you were using, then run everything else within that simulation. And so on. The pattern is that the very last stage of the work is taken care of by the universe itself, and the more of the universe's work you choose to do yourself, the harder your life will be. The universe drags the ball down in the first case, runs the computer in the second case, and runs the second computer in the third case.
Cont...

>> No.1088891

>>1088885
Let's apply that to brains. To simulate a brain, you don't need merely the computational power of a brain. You need the computational power requisite to run to simulate the brain itself (a real brain does not have to do this; the brain does not have to "think" about the positions of its own ions and whatnot), and ON TOP of that, you have to simulate how the brain reacts and responds to input. This makes the task a lot more complicated.

There is a question of native versus non-native running of an application in computing very frequently. For example, you may know that a lot of mobile devices don't support Flash, because it requires a lot of computing power and electrical power to do so. The CPU has to simulate a Flash environment, then run that simulation with inputs. One possible workaround is to build a hardware native Flash chip on the device. This would be a hardware Flash environment. The CPU would no longer have to simulate a Flash environment, saving huge amounts of electrical and computing power, and would only have to give the input, power the chip, and take the output. This explains my suggested alternative to simulation, which is creating a computer with a matching architecture, i.e. a neural network.

>> No.1088987

>>1088885
>>1088891
ahhhww all that typing and not even one response

>> No.1088993

lol

>> No.1089017
File: 16 KB, 377x441, 1275399162144.jpg [View same] [iqdb] [saucenao] [google]
1089017

>>1088987 Because the front page is needed to debate whether Jesus is real or not for the hundredth time today

>> No.1089025 [DELETED] 

>>1088891
>Implying simulating the human brain will work like a emulator.

>> No.1089047

>>1089025
No, I'm not implying that. You must not have followed the conversation from the start, because those long posts are replying to someone who thinks that. I am showing it to be infeasible.

>> No.1089056

>>1088659
the links have sources/references.. some of which are journals.. can't you click/read through them before opining?

>>1088891
>To simulate a brain
you simply do not need to simulate a brain to create something intelligent.

>> No.1089060

Its optimistic I would say.

>> No.1089062

>>1089056
> you simply do not need to simulate a brain to create something intelligent
cool speculation bro

>> No.1089111

>>1089062
simulation is one method. just creating something intelligent is another.

if neurons and axons were legos, you could put them together in a way to make something intelligent.

if electrons and transistors were legos, you could put them together in a way to make something intelligent.

there is nothing special about a biological substrate. saying intelligence can only exist in the way evolution/biology has created it is like saying animal-looking and animal-functioning things can only come from biology, not metal or steel or robotics. the only argument against straight creating intelligence on a computer would be something from religion/god/ whatever.

http://www.youtube.com/watch?v=W1czBcnX1Ww

>> No.1089129

>>1089056
I'm sure we'll eventually create AI in more efficient ways in the future, but for now brains are our best bet. We haven't even got any evidence that intelligence can exist on current supercomputers without heavy emulation, so the idea that we just need some software is extremely speculative.

>> No.1089146

you suck

>> No.1089151

>>1089111
>if electrons and transistors were legos, you could put them together in a way to make something intelligent.
>different architecture

The same as what I said, then...

>> No.1089179

>>1089111
I'd love to agree or disagree with your position, but it is so vague as to be Matrix-quality speculation.

>> No.1089222

Shit rec & play, have you been here the whole time?

Any dualists show up?

>> No.1089227

>>1089222
Don't turn this into a tripfag circlejerk goddamn it.

>> No.1089282

>>1089179
My position:
Intelligent systems can exist on computers.

Reality:
Intelligent systems exist on biological substrates.

What's there to disagree with?

>> No.1089766
File: 23 KB, 471x355, Shit Nigga.jpg [View same] [iqdb] [saucenao] [google]
1089766

>>1089222
No fizx, don't think they're smart enough to debate about AI.

>>1089227
Don't be jealous, you can join in! Just fill in that lovely little name field. ~