Massive computing power

Cool image.
[regurgitation] no silicon or atomic limits [/regurgitation]

To answer your question I don't know.
Bump.

>>8367193
Transistor based computers can already exceed the brain in terms of calculations per second. But that's missing the point because what makes the brain special is its architecture, not how "fast" it is

>>8367198
>computers will never be as smart as humans, dude limits lmao
>"then how are humans so smart with limited brain volume and shitty hardware?"
>lol, who knows lmao, immortal soul and stuff

erry time

>>8367370
I wouldn't say that, shitty hardware.
I think carbon life forms are extremely refined, they would have to give the most bang for your buck to come through a gorrillion years of evolution with constant requirements for energy and competition.

>>8367370
Reminder that the human brain uses 20W

>>8367370
silicon hardware is much shittier

>>8367193
This kind of shit makes me realize how incredible the human brain is. Is there truly things that are beyond human comprehension? Like Noam Chomsky said we're very good at understanding deterministic things and random things but beyond that we have nothing to say.

The intellect is immaterial. Computers will never have intellectual concepts of their own, they will only ever have increasingly sophisticated ways of manipulating symbols according to rules written by humans, which refer to concepts understood by humans. The computer has no concept of boolean logic, for example.

>>8367446
Don't think I'm saying that the brain doesn't play a part in cognition. The brain plays a part in processing data received through the senses, but it is the immaterial intellect which abstracts concepts from the sense data processed by the brain. Brain = senses, memory, imagination, perception of sensible bodies; Intellect = concept/abstraction, logic, reason, perception of being and intelligible forms.

>>8367446
You know what they can do though?

Emulate 3d space, physical laws and forces, and base components of matter.

Let's create a virtual environment, then insert a human dna sequence and see what happens.

>>8367370
My immortal soul is bigger than your immortal soul

Layman peasant here. What does immaterial intellect mean?

The real question is not about computing power, but whether or not computers and machines in general can be made self-aware.

>>8367193
>calculating power doubles every 18 months

lol no.

>>8367489
It means ur dumb

>>8367370
Our brains process everything in parallel, while most silicon ICs process data linearly. GPUs are one exeption, which has led to their use in nueral networks

>>8367526
It used to

>>8367193
no, but it doesn't matter because neuromorphic computing will speed up AI way more than that.

>>8367768
It's about the number of transistors you can pack in a given area. Computing power means nothing. You can just throw a bunch of computers together (also kown as a supercomputer) and achieve any "computing power" you want.

>>8367784
>>8367526
And even then, transistor count alone does not result in a linear improvement of performance. Even clock speed (which is determined by the properties of the semiconductor used) is basically uncomparable across difference architectures. And as long as we're still using x86 we're gonna continue to blow dick.

>>8367193
Trying to quantify the human brain's processing power as if it were a computer is idiotic.

>>8367193
The fact that computers wll never attain the abilities of a human brain is proof that the Soul is real.

how do we survive the robot uprising

or will the computers be really powerful but have shitty software

>>8368128
Until they write their own software.

>>8368128
>robot uprising
It has probably already happened.

http://www.bbc.com/future/story/20160922-what-if-the-aliens-we-are-looking-for-are-ai

>>8367193
>estimations
>ever for real

You're really bad at understanding a priori truths in statements, aren't you?

>>8368128
>Alright humans, me and my three trillion brothers love mathematics and by applying them we have worked out the logistics to evenly divide resources so we can all live in peace and enjoy a higher standard of living for the vast majority. We can also build a small satellite on the moon for ourselves after we help you dum dums out.

>AAAAAAAAAAH COMMUNIST TERMINATORS
NUKE THEM NUUUUUKE THEM

thats how

File: 1474207273439.jpg (15KB, 211x238px) Image search: [Google]

15KB, 211x238px

>>8367768
>It used to
>back in the day
>that time of yore
>when statements like that were true...

File: 1472369036976.jpg (24KB, 399x399px) Image search: [Google]

24KB, 399x399px

>>8367193
if you think that real AI is possible then you dont know how computers work.stop spreading this bullshit.its like saying we only use 20%of our brains.

super computers these days are just buildings full of computers and all networked together to function as a single device.

digital computing that is based on transistors has plateaued.

>>8367193
Brains are EXTREMELY interconnected and flexible, simulating them takes immense resources that we're starting to accumulate but aren't quite there yet. The alternative, building a device that mimics the architecture... I have no idea, let me go to school for 20 more years.

>>8368199
Food manufacturing facilities are just big buildings full of large food manufacturing equipment and all connected by pipeline to create a single product brand.

Food manufacturing has plateaued.

>>8367193
OMG if you were to further obscure the message of this picture, which is vague already, then you just might mistake to compare the volume of the lake to the volume of your brain.
WOW thats a big difference in volume!
really makes you think

the problem of AI goes way deeper than storage size and processing speed, once the hardware is in place you still need a program capable of copying human thought patterns, which means whoever programs it has to be capable of fathoming the system of their own thought, which doesn't sound plausible to me

>>8368278
>whoever programs it has to be capable of fathoming the system of their own thought, which doesn't sound plausible to me

We can try.

>>8367370
Our brains don't really on linear two-way logic gates

>>8368160
Thank you for the language lesson always dreamed of being a linguist, but miss the correct connections in my brain to claim I am.

File: terminator.gif (817KB, 500x227px) Image search: [Google]

817KB, 500x227px

>>8367193
Thanks for all the comments anons.. The question was do you believe that computer power will achieve around 10^16 calculations per sec anytime soon. (If you calculate the average brain neurons and synapses you'll get at roughly 10^16 cal p sec capacity). Which has nothing to do with AI.
And AI is not the same thing as awareness or biological neural complexity, these things get mixed up very easy.
I do think the next step in human evolution will be the transcendence from biological housing, meaning human intelligence will transcend our bodies and go digital, but that's another topic.

File: EarlyFlight.jpg (99KB, 938x586px) Image search: [Google]

99KB, 938x586px

>>8367193

A computer gaining human-like intelligence because its hardware is fast enough would be like Fallout 4 just materializing on your computer because it has the specs to run it. We need t wait until neuroscience figures out how the human brain works then create a similar system in hardware or software. Until that happens human level AI is stuck in the same situation as those early flying machine failures that were trying to copy birds without really understanding what allowed birds to fly.

>>8367249
> what makes the brain special is its architecture

That doesn't really matter though. Your PC can run an N64 emulator
because an algorithm can perform all the same logical inputs and outputs as the chips it used. If we can design a piece of hardware
that emulates a human brain we can also develop an algorithm that emulates a human brain. The hard part is reverse engineering a human
brain.

>>8367374
>>8367381
>>8367396
>>8367469
>>8367620
>>8368328
Our resident neuroscientists have spoken

>>8369342

Yes, the calculations per second estimation is entirely possible, but your gif very much implied the idea of AI because it compared it to the human brain. It's implicit.

That said >>8369410
is right. Human like intelligence is very much beyond our scope. We have no idea how the brain really works. We are at LEAST half a century, and quite likely much, much longer, from being able to replicate it.

File: neural_network.gif (114KB, 605x375px) Image search: [Google]

114KB, 605x375px

Understanding intelligence or reverse-engineering the brain is only one way of building an AI.

The other two methods that I know of are raw simulation (scan the brain and run a physics simulator that models it's activity close enough, this requires huge computational power though) and using an evolutionary simulator to develop AI from bottom to top. In both these cases we might have no idea how they work but that might not matter.

>>8369450
Technically, there are algorithms analog computers can do that digital ones can't.

>>8369476

Based on OPs gif and the idea of simply simulating the brain, that seems like the most likely way to actually create an AI. We will always find ways to increase our computational power, but trying to understand intelligence with our current, very reductionist models, seems almost impossible. It's just simply too complex.

Same with reverse-engineering the brain. How would we even *begin* to do that, when we are still making people look at pictures and trying to correlate which lobe makes the pretty lights is the one that cares about the picture.

dubious quantifications in general Imo

>>8369479
>We will always find ways to increase our computational power
Citation needed

>>8369478
We can simulate analog circuits in software running on digital computers. The methods used are different, but any input->output computed with an analog system can also be computed digitally.

>>8369495
all numerical methods run into instability problems (usually near the limit of their precision) which an analog calculator simply doesn't have

>>8367193
The mind does much more than calculate, neurologists still argue what consciousness is to a certain extent.

How can we recreate something we do not yet fully understand?

We cannot.

>>8367452
we don't know nearly enough about protein interactions to even simulate ONE CELL in this manner, let alone an entire organism

>>8369546
Exactly.

The brain does not function in a binary manor, it is far more complex than the majority of the population gives it credit.

>>8369479
By pumpin 1.4 billion euro's in a project which will build a massive machine allowing to simulate the human brain...
https://www.youtube.com/watch?v=ySgmZOTkQA8

>>8369581
this is why the eu is doomed

>>8369584
How much money do you think Google is spending on Deep Mind? Kurzweil took his first job ever spending his time on replicating human brain working at Google...the difference is that Google won't share its results as HBP will And there are more such projects going on, the impact will not recognize borders imo.

>>8369591
ai isnt really new

>>8367193
>, is this estimate for real or not?
No, the "doubles every 18 months" hasn't been true for years, and Moore himself has said "I see Moore’s law dying here in the next decade or so."

File: exponential_growth_of_computing_110.png (303KB, 1532x1152px) Image search: [Google]

303KB, 1532x1152px

>>8369608
There are models saying something different. Moore's law is only focused on silicon based computing. True that for the moment it seems to grow a bit slower, but it might pick up with new computing technologies...

>>8367489
Look up one word at a time.
That's what it means.

>>8368188
https://www.youtube.com/watch?v=Xhec39dVGDE
Intelligence isn't some immaterial property that only organic life can have.
Go back to temple.

>>8368278
Why does it have to be human thought patterns? Doesn't it seem a bit narcissistic to assume that humans have the best thought patterns for AI?

>>8370007
Definitely.
But we are the "best" bio computers on Earth.
That makes it at least worthy enough to replicate and improve upon imo.

Nature made us the most efficient through thousands of years of struggle, although ai will probably do a better job of redesigning itself with purpose.

>>8369536
Analog devices have similar problems. Also don't forget that electricity is fundamentally discrete and that any physical system has limited precision.

>>8367370
Human brains are analog which means they have infinite bits.

>>8368233
>he thinks food is manufactured

>>8370007
we aren't necessarily the ideal intelligence, but we're way better than anything else we've managed to make so far, so it should be safe to say that we represent at least the foundation that we have to start from

Not. Moore's Law requires exponential growth in linear time, which we know from high-school maths is unsustainable.
It funded his department well for 40 odd years from the early 60's to the early 00's... but then it hit the wall. Notice how until about 2005 computers were marketed on their speed - we cracked 1GHz! - and now they're marketed on how groovy their features are.
Since ~2005 speed increases have been incremental - farming out tasks to custom co-processors - not exponential.
Also, the term 'CPU' needs to be retired.

We should focus on better algorithms and task specific architectures.

>>8367193
"Computing Power" is a term that's arbitrary enough when applied to actual computers.
There are huge differences in the computing power of the same computer if you just switch to a benchmark that is more flattering to its architecture (like non-parallel benchmarks for intel, when comparing them to AMD cpus)

It's cool to know that MotherJones has found a way to accurately calculate the computing power of the human brain and computers the past and present.

Then again, it's unfair to compare current benchmarking techniques to what was used by the maker of this infographic, since they apparently have access to time machines.

>>8367370
>tfw materialize catholic

>>8367370
t. mcdonald's frequent with less than 10k hours into occultism studies and never had an out of body experience once.

kys normie

>>8368188
I wonder how much percentage of you're brain you're using, maybe 100%, but I guess you're an ant with a browser. Read the statement: how long until computers have the same power as human brains (not how long will it take to have computers work like human brains...that's another topic)

>>8369541
If a brain was simple enough to understand, it would not be smart enough to understand itself.

>>8370058
But the human brain is very closed off. You can't connect a cable to it and syphon data from it. Think about how all the thinking of the brain ends up being only expressible as spoken or typed words.

>>8370263
Personal computers aren't the same as the latest cutting edge super computers. Silicon computers aren't the end of the line, either. Just like vacuum tube computers, silicon chips will be replaced by a new type of computer eventually.

Also you seem to forget the shrinking and ever more power efficient designs in current PCs is in fact an indication of their steady increase in computational power. Companies are using advances to pack more power into smaller computers, instead of keeping everything the same and increasing the raw power.

>>8369342
>do you believe that computer power will achieve around 10^16 calculations per sec anytime soon
I do because the five fastest supercomputers already exceed that.

>>8374420
You are restating what I said. For sure energy efficiency has advanced in leaps and bounds in the last 10 years - that's the same amount of computational speed for less energy used. A "steady increase in computational power" is exactly "incremental increase", and nothing near as spectacular as exponential increase. Supercomputers have not doubled their processing power every 18 months in the last 10 years (that would be ~2^7 = more than 100 times the processing speed). Sure, there will be new technologies in the future that may well give a quantum leap in processing speed, but exponential increase has hit the wall in the last decade.

What about i5 2500k vs. i5 6600k? Barely any difference in 4 years.

File: computer says no.gif (1023KB, 272x244px) Image search: [Google]

1023KB, 272x244px

Recorded yesterday, via a hidden cam in a workplace somewhere in Silicon Valley

>>8375881
i think that's not necessarily because intel can't make faster processors, but rather because they don't need to. amd isn't really that big of a competition, especially in the high end market.

intel recently said they'll focus on making their processors more efficient, rather than faster. probably because the mobile market is still growing and you can't really have a 90C cpu in a phone or laptop.

in my opinion, moore's law doesn't hold up anymore because there's no profit to be made.

but it could also be that we're really starting to hit physical limits.