Peak Oil
Donate Bitcoin

Donate Paypal


PeakOil is You

PeakOil is You

"Bluebrain | Year One" by Couple 3 Films

A forum to either submit your own review of a book, video or audio interview, or to post reviews by others.
Post a reply

"Bluebrain | Year One" by Couple 3 Films

Unread postby Carlhole » Sat 27 Feb 2010, 05:21:05

Image

vimeo

$this->bbcode_second_pass_quote('', 'H')enry Markram is attempting to reverse engineer an entire human brain, one neuron at a time. This piece is an introduction to director Noah Hutton's 10-year film-in-the-making that will chronicle the development of The Blue Brain Project, a landmark endeavor in modern neuroscience.


Alright, a documentary! The trailer looks good. Hopefully, I won't have to wait forever for it to appear.

You know, folks, we're going to see machine intelligence transform the world if we just manage to live another decade or two. It will probably happen at the same time the world is ramping up its fusion energy capacity.
Carlhole
 
Top

Re: "Bluebrain | Year One" by Couple 3 Films

Unread postby Dezakin » Sat 27 Feb 2010, 06:29:35

$this->bbcode_second_pass_quote('Carlhole', 'Y')ou know, folks, we're going to see machine intelligence transform the world if we just manage to live another decade or two. It will probably happen at the same time the world is ramping up its fusion energy capacity.

Incredibly unlikely. The task to simulate human brain processes is a large puzzle all on its own, likely requiring a couple of decades before you have something that is similar enough to learn the way a human does. Then you have to actually raise it, like a child, which from past experience, requires another couple of decades. Assuming very fast track development this gives you a vastly different world in perhaps 40 years compared to today, but no sooner. This is interesting work and its part of the work that will transform all of civilization, but not that fast. It will most likely be useful for developing drugs in the short term, and AI applications sometime in the next 50 years.

As for fusion, unless there is a vastly different technology that has vastly lower capital requirements, it will never compete with fission power. Polywell would do it if they actually managed to satisfy all the claims they make, but its unclear and unlikely that it actually works.
User avatar
Dezakin
Heavy Crude
Heavy Crude
 
Posts: 1569
Joined: Wed 09 Feb 2005, 04:00:00
Top

Re: "Bluebrain | Year One" by Couple 3 Films

Unread postby Human Brain 2045 » Sat 27 Feb 2010, 12:57:52

Check this 2 Astonishing lectures of Henry Markram:

1. http://neuroinformatics2008.org/congres ... m.flv/view

2. http://ditwww.epfl.ch/cgi-perl/EPFLTV/h ... in_close=0


Very interesting article from the ‘Seed’ magazine:

http://seedmagazine.com/content/article/out_of_the_blue

http://seedmagazine.com/content/article ... he_blue/P2


And a few more interesting links about the project:

http://news.bbc.co.uk/2/hi/sci/tech/8012496.stm

http://www.youtube.com/watch?v=Bz5IUaRr8No

http://www.youtube.com/watch?v=RLCT3wU4fek
Human Brain 2045
Wood
Wood
 
Posts: 1
Joined: Sat 27 Feb 2010, 12:52:51
Top

Re: "Bluebrain | Year One" by Couple 3 Films

Unread postby TWilliam » Sat 27 Feb 2010, 13:19:14

$this->bbcode_second_pass_quote('Carlhole', 'Y')ou know, folks, we're going to see machine intelligence transform the world if we just manage to live another decade or two.

And how!

Image
"It means buckle your seatbelt, Dorothy, because Kansas? Is goin' bye-bye... "
User avatar
TWilliam
Expert
Expert
 
Posts: 2591
Joined: Sun 28 Nov 2004, 04:00:00
Top

Re: "Bluebrain | Year One" by Couple 3 Films

Unread postby Carlhole » Sat 27 Feb 2010, 14:23:51

$this->bbcode_second_pass_quote('pstarr', 'T')he Singularity prefers lattes...


Trying posting something interesting for a change, pstarr. You might enjoy it.

Do you think that it is possible and routine for scientists to learn about biology? Yes?

Well, then, they will learn about the brain's biology. And then they will apply those principles. The more they learn, they more they will apply. And that simple process leads to machine intelligence - which may or may not resemble human intelligence, but will certainly entail a machine "mind" manipulating complex, fluid symbols and concepts.

Nature already did it. It's proven!

In order for your position to be credible, you have to posit that there will be complex features in our regular world (physics or biology) that will remain hidden to us simply because we are not smart enough to figure them out, despite all our best efforts.

But that is not what our experience has shown. Experience shows that human beings wrestle with advanced problems for ages - wondering... creating philosophy, theories, etc. But then, when key technological pieces fall in place (like supercomputers, nano-microscopy, etc.), sudden huge advances occur which change everything.

There is ZERO indication that there is some sort of intellectual "wall" sealing us off from capabilities every bit as stunning as what occurs in Nature. Once features as complex as conscious machine brains are able to exist in supercomputers, what do you think that will do to the pace of evolution within those mediums?

Evolutionary processes are every bit as workable in a virtual realm as in the physical world - except that you don't have to carry a lot of real, meat, ass around with you and you're not dependent on real time, and that sort of thing. So evolution would take place extremely rapidly.

That's what the Singularity is.
Carlhole
 
Top

Re: "Bluebrain | Year One" by Couple 3 Films

Unread postby EnergyUnlimited » Sun 28 Feb 2010, 04:23:57

$this->bbcode_second_pass_quote('Carlhole', '
')Well, then, they will learn about the brain's biology. And then they will apply those principles. The more they learn, they more they will apply.

Learning about some process not necessarily imply ability to replicate capability of actual thing.
$this->bbcode_second_pass_quote('', 'A')nd that simple process leads to machine intelligence - which may or may not resemble human intelligence, but will certainly entail a machine "mind" manipulating complex, fluid symbols and concepts.

It may also lead to nowhere and die out once budget for the project is exhausted.

$this->bbcode_second_pass_quote('', 'N')ature already did it. It's proven!

And Natural product doesn't work base on silicon, doesn't employ any sort of digital process to perform it's tasks and it's information storage and memory handling is vastly different than one available to electronic engineers.
$this->bbcode_second_pass_quote('', 'B')ut that is not what our experience has shown. Experience shows that human beings wrestle with advanced problems for ages - wondering... creating philosophy, theories, etc.

Performance of the past is not necessarily a guideline for performance of the future.
Financial markets have learned it and science is next in line.
And you know, Kurzweil will die one day, even if he doesn't want to...
$this->bbcode_second_pass_quote('', 'B')ut then, when key technological pieces fall in place (like supercomputers, nano-microscopy, etc.), sudden huge advances occur which change everything.
And due to resources related problems prices of various pieces in all probabilities are going to increase.

$this->bbcode_second_pass_quote('', 'T')here is ZERO indication that there is some sort of intellectual "wall" sealing us off from capabilities every bit as stunning as what occurs in Nature.
Window related to resource availability is just shutting down.
$this->bbcode_second_pass_quote('', 'O')nce features as complex as conscious machine brains are able to exist in supercomputers...
And this is not going to happen.
Human brain works on vastly different principles than digital technology.
You cannot make a coffee from rat droppings, you know...

$this->bbcode_second_pass_quote('', 'E')volutionary processes are every bit as workable in a virtual realm as in the physical world - except that you don't have to carry a lot of real, meat, ass around with you and you're not dependent on real time, and that sort of thing.
Such process (if achieved) would constantly call for hardware upgrades, so overall progress would be slow due to difficulties with these.
$this->bbcode_second_pass_quote('', 'S')o evolution would take place extremely rapidly.
Not at all.
Necessity of hardware upgrades would slow it down and bring to halt once physical limits related to components size/performance are reached and in all probabilities much faster due to countless engineering problems.
User avatar
EnergyUnlimited
Light Sweet Crude
Light Sweet Crude
 
Posts: 7537
Joined: Mon 15 May 2006, 03:00:00
Top

Re: "Bluebrain | Year One" by Couple 3 Films

Unread postby Cog » Sun 28 Feb 2010, 05:11:56

$this->bbcode_second_pass_quote('Dezakin', '')$this->bbcode_second_pass_quote('Carlhole', 'Y')ou know, folks, we're going to see machine intelligence transform the world if we just manage to live another decade or two. It will probably happen at the same time the world is ramping up its fusion energy capacity.

Incredibly unlikely. The task to simulate human brain processes is a large puzzle all on its own, likely requiring a couple of decades before you have something that is similar enough to learn the way a human does. Then you have to actually raise it, like a child, which from past experience, requires another couple of decades. Assuming very fast track development this gives you a vastly different world in perhaps 40 years compared to today, but no sooner. This is interesting work and its part of the work that will transform all of civilization, but not that fast. It will most likely be useful for developing drugs in the short term, and AI applications sometime in the next 50 years.

As for fusion, unless there is a vastly different technology that has vastly lower capital requirements, it will never compete with fission power. Polywell would do it if they actually managed to satisfy all the claims they make, but its unclear and unlikely that it actually works.


I disagree with part of your statement. In order to raise this electronic child, we don't need 20 years to let it learn as we do as humans. The sum of human knowledge could be uploaded to this artificial brain. A silicon based brain, if developed this way, would be far different then the human analogue. Machines could be programmed to show emotion but they won't feel it.
User avatar
Cog
Fusion
Fusion
 
Posts: 13416
Joined: Sat 17 May 2008, 03:00:00
Location: Northern Kekistan
Top

Re: "Bluebrain | Year One" by Couple 3 Films

Unread postby Homesteader » Sun 28 Feb 2010, 07:28:41

$this->bbcode_second_pass_quote('Cog', ' ')Machines could be programmed to show emotion but they won't feel it.


Sounds like my ex. Cr*P! You don't suppose. . . .. .?
"The era of procrastination, of half-measures, of soothing and baffling expedients, of delays, is coming to a close. In its place we are entering a period of consequences…"
Sir Winston Churchill

Beliefs are what people fall back on when the facts make them uncomfortable.
User avatar
Homesteader
Heavy Crude
Heavy Crude
 
Posts: 1856
Joined: Thu 12 Apr 2007, 03:00:00
Location: Economic Nomad
Top

Re: "Bluebrain | Year One" by Couple 3 Films

Unread postby TWilliam » Sun 28 Feb 2010, 14:18:08

Once one comprehends that man is more than simply the sum of his parts, one will likewise understand why machine 'consciousness' will not be happening anytime soon. Our self-awareness is not just an epiphenomenon of electro-chemical processes. When all is said and done, a computer is still just an adding machine. No matter how fast you make an adding machine, no matter how many calculations per second it's able to perform, that's all it will ever be...
"It means buckle your seatbelt, Dorothy, because Kansas? Is goin' bye-bye... "
User avatar
TWilliam
Expert
Expert
 
Posts: 2591
Joined: Sun 28 Nov 2004, 04:00:00
Top

Re: "Bluebrain | Year One" by Couple 3 Films

Unread postby Dezakin » Mon 01 Mar 2010, 00:54:28

$this->bbcode_second_pass_quote('Cog', '')$this->bbcode_second_pass_quote('Dezakin', '')$this->bbcode_second_pass_quote('Carlhole', 'Y')ou know, folks, we're going to see machine intelligence transform the world if we just manage to live another decade or two. It will probably happen at the same time the world is ramping up its fusion energy capacity.

Incredibly unlikely. The task to simulate human brain processes is a large puzzle all on its own, likely requiring a couple of decades before you have something that is similar enough to learn the way a human does. Then you have to actually raise it, like a child, which from past experience, requires another couple of decades. Assuming very fast track development this gives you a vastly different world in perhaps 40 years compared to today, but no sooner. This is interesting work and its part of the work that will transform all of civilization, but not that fast. It will most likely be useful for developing drugs in the short term, and AI applications sometime in the next 50 years.

As for fusion, unless there is a vastly different technology that has vastly lower capital requirements, it will never compete with fission power. Polywell would do it if they actually managed to satisfy all the claims they make, but its unclear and unlikely that it actually works.


I disagree with part of your statement. In order to raise this electronic child, we don't need 20 years to let it learn as we do as humans. The sum of human knowledge could be uploaded to this artificial brain. A silicon based brain, if developed this way, would be far different then the human analogue. Machines could be programmed to show emotion but they won't feel it.

Oh sure, we could develop an entirely artificial intelligence that would be vastly different from human intelligence. But that would require far more work than copying an existing design. Given what we have, we'd have to train it the same way you train humans. After you train it you can make as many copies as you want, but its not a simple matter of uploading information.
User avatar
Dezakin
Heavy Crude
Heavy Crude
 
Posts: 1569
Joined: Wed 09 Feb 2005, 04:00:00
Top

Re: "Bluebrain | Year One" by Couple 3 Films

Unread postby Dezakin » Mon 01 Mar 2010, 00:58:07

$this->bbcode_second_pass_quote('TWilliam', 'O')nce one comprehends that man is more than simply the sum of his parts, one will likewise understand why machine 'consciousness' will not be happening anytime soon. Our self-awareness is not just an epiphenomenon of electro-chemical processes. When all is said and done, a computer is still just an adding machine. No matter how fast you make an adding machine, no matter how many calculations per second it's able to perform, that's all it will ever be...

Honestly this is epistimilogical wankery. We'll never know anything is self aware. I cant know that any other human isn't just an adding machine. Who cares. We only care about behavior.

Replicating a brain in software is one path to this, and weather or not you think its conscious is immaterial to the applications.
User avatar
Dezakin
Heavy Crude
Heavy Crude
 
Posts: 1569
Joined: Wed 09 Feb 2005, 04:00:00
Top

Re: "Bluebrain | Year One" by Couple 3 Films

Unread postby Carlhole » Mon 01 Mar 2010, 01:04:40

$this->bbcode_second_pass_quote('Dezakin', '')$this->bbcode_second_pass_quote('TWilliam', 'O')nce one comprehends that man is more than simply the sum of his parts, one will likewise understand why machine 'consciousness' will not be happening anytime soon. Our self-awareness is not just an epiphenomenon of electro-chemical processes. When all is said and done, a computer is still just an adding machine. No matter how fast you make an adding machine, no matter how many calculations per second it's able to perform, that's all it will ever be...

Honestly this is epistimilogical wankery. We'll never know anything is self aware. I cant know that any other human isn't just an adding machine. Who cares. We only care about behavior.

Replicating a brain in software is one path to this, and weather or not you think its conscious is immaterial to the applications.


A lot of people argue that WE aren't really conscious either.
Carlhole
 
Top

Re: "Bluebrain | Year One" by Couple 3 Films

Unread postby TWilliam » Mon 01 Mar 2010, 02:49:08

$this->bbcode_second_pass_quote('Dezakin', 'H')onestly this is epistimilogical wankery. We'll never know anything is self aware. I cant know that any other human isn't just an adding machine. Who cares. We only care about behavior.

Replicating a brain in software is one path to this, and weather or not you think its conscious is immaterial to the applications.

The only wankery with regard to this particular issue resides within the assertion that self-awareness doesn't exist. If it didn't exist, there would be no one to make the assertion. What utter idiocy for a mind to deny its own reality. The nature of mind may be endlessly debatable, but its reality is self-evident..

Objectivism is useful to a point, but it is not the totality of existence. One might be able to model behavior, but that is an entirely different thing than reproducing consciousness.
"It means buckle your seatbelt, Dorothy, because Kansas? Is goin' bye-bye... "
User avatar
TWilliam
Expert
Expert
 
Posts: 2591
Joined: Sun 28 Nov 2004, 04:00:00
Top

Re: "Bluebrain | Year One" by Couple 3 Films

Unread postby Dezakin » Mon 01 Mar 2010, 03:34:05

$this->bbcode_second_pass_quote('TWilliam', '')$this->bbcode_second_pass_quote('Dezakin', 'H')onestly this is epistimilogical wankery. We'll never know anything is self aware. I cant know that any other human isn't just an adding machine. Who cares. We only care about behavior.

Replicating a brain in software is one path to this, and weather or not you think its conscious is immaterial to the applications.

The only wankery with regard to this particular issue resides within the assertion that self-awareness doesn't exist. If it didn't exist, there would be no one to make the assertion. What utter idiocy for a mind to deny its own reality. The nature of mind may be endlessly debatable, but its reality is self-evident..

Objectivism is useful to a point, but it is not the totality of existence. One might be able to model behavior, but that is an entirely different thing than reproducing consciousness.

If you model behavior of the brain, it will produce the same behavior. A model of a calculator is as good as a calculator. Who cares if it reproduces consciousness. I cant tell if you're conscious anymore than you can tell I am. What are you trying to say, because you aren't being very clear about it.
User avatar
Dezakin
Heavy Crude
Heavy Crude
 
Posts: 1569
Joined: Wed 09 Feb 2005, 04:00:00
Top

Re: "Bluebrain | Year One" by Couple 3 Films

Unread postby EnergyUnlimited » Mon 01 Mar 2010, 03:58:20

Dezakin,
Models are only imperfect approximations of actual thing.
In particular no amount of digital technology can model human brain to any accurate degree because working of such brain is not digital even in principle and brain is not even using any sort of binary code in its memory/information processing.

So it may be possible to create some AI using digital technology but nature of working of such AI will not resemble workings of human brain to any significant degree.

So those who want to get some more accurate human brain pretending AI must abandon digital technology and start new branch of computing science from the scratch.
I do not see it happening yet.
User avatar
EnergyUnlimited
Light Sweet Crude
Light Sweet Crude
 
Posts: 7537
Joined: Mon 15 May 2006, 03:00:00
Top

Re: "Bluebrain | Year One" by Couple 3 Films

Unread postby Dezakin » Mon 01 Mar 2010, 05:17:31

$this->bbcode_second_pass_quote('EnergyUnlimited', 'D')ezakin,
Models are only imperfect approximations of actual thing.
In particular no amount of digital technology can model human brain to any accurate degree because working of such brain is not digital even in principle and brain is not even using any sort of binary code in its memory/information processing.

This is a common objection based on the highly implausible notion that a perfect model of the molecular interactions are necessary to produce similar behavior to observed biological nervous systems. Experience so far with virtual nematodes and mammalian neocortical columns do not indicate that perfect models are necessary to replicate behavior. We don't need to do ab initio modeling to produce neurological modeling with similar behavior based on all the results so far and we have no reason to expect that going to larger models this will change. More likely we'll be able to reduce to simpler, more imperfect models as we find what is and isn't necessary for producing similar systems. And sure, that's a long road ahead, possibly decades.

$this->bbcode_second_pass_quote('', 'S')o those who want to get some more accurate human brain pretending AI must abandon digital technology and start new branch of computing science from the scratch.
I do not see it happening yet.

You can't do anything on an analog machine that you can't do on a digital machine in practice.
User avatar
Dezakin
Heavy Crude
Heavy Crude
 
Posts: 1569
Joined: Wed 09 Feb 2005, 04:00:00
Top

Re: "Bluebrain | Year One" by Couple 3 Films

Unread postby EnergyUnlimited » Mon 01 Mar 2010, 08:20:05

So are those models of neocortical columns doing anything useful?
Or may be they are merely pretending some sort of electrical patterns present in actual neocortical column but do not display any useful functions above that.

$this->bbcode_second_pass_quote('Dezakin', 'Y')ou can't do anything on an analog machine that you can't do on a digital machine in practice.

The most important features which are going to be lost in digital systems are related to creativity.
It is degree of randomness and related issues of handling error (quite rampant in analogue devices) what has a lot to do with creativity of brain.
Creativity appears to come as a product of countless errors of quite fine tuned magnitude, followed by error correction mechanisms (which are imperfect), selecting more promising outcomes for future processing (and we have no idea, how this selection works...) and after a number of iterations of such process some useful creative output is presented at the end.
Analogue devices are more suitable for such tasks.
On the other hand digital devices are working base on rubbish in = rubbish out principle.
No creativity there, just mere bit processing.

In any case I am not seeing any artificial brains around so certainly digital technology is not providing these and IMO it never will, so up to date analogue one is definitely superior in this respect.
.

Didn't you think why our genetic code appears to be grossly misdesigned from perspective of information theory?
It is inefficient, not binary - means uneconomic from data storage perspective, prone to countless translation errors, spontaneous and induced mutations etc.
However these are particular features which are allowing for species evolution and because of them life is creative.

So make an information processing system too perfect and it will merely deliver input - output processing, but show no ingenuity or problem solving ability.
That is a feature of digital systems.
So you have to work on idea, how to make your brain model sufficiently crap and error prone to make ingenuity possible and yet sufficiently robust to prevent calculations diverging into loads of rubbish.
There might be an extreme fine tuning required here, perhaps impossible to achieve in digital systems.
User avatar
EnergyUnlimited
Light Sweet Crude
Light Sweet Crude
 
Posts: 7537
Joined: Mon 15 May 2006, 03:00:00
Top

Re: "Bluebrain | Year One" by Couple 3 Films

Unread postby Dezakin » Mon 01 Mar 2010, 13:30:12

$this->bbcode_second_pass_quote('EnergyUnlimited', 'S')o are those models of neocortical columns doing anything useful?
Or may be they are merely pretending some sort of electrical patterns present in actual neocortical column but do not display any useful functions above that.

I can't speak for the work on the neocortical columns except that the neurologists working on the project seem to be satisfied with the results. However, full simulation of simpler organisms, c elegans nervous system performs in the same way as the real thing, seeking food and the like.

$this->bbcode_second_pass_quote('EnergyUnlimited', '')$this->bbcode_second_pass_quote('Dezakin', 'Y')ou can't do anything on an analog machine that you can't do on a digital machine in practice.

The most important features which are going to be lost in digital systems are related to creativity.
It is degree of randomness and related issues of handling error (quite rampant in analogue devices) what has a lot to do with creativity of brain.
Creativity appears to come as a product of countless errors of quite fine tuned magnitude, followed by error correction mechanisms (which are imperfect), selecting more promising outcomes for future processing (and we have no idea, how this selection works...) and after a number of iterations of such process some useful creative output is presented at the end.
Analogue devices are more suitable for such tasks.

You're missing some basic concepts of information theory. You can model any analog system on a digital system to however many orders of magnitude as you like. There isn't any inherent advantage to running things on analog systems from an information theoretic view, and really are often at a disadvantage.

$this->bbcode_second_pass_quote('', 'O')n the other hand digital devices are working base on rubbish in = rubbish out principle.
No creativity there, just mere bit processing.

In any case I am not seeing any artificial brains around so certainly digital technology is not providing these and IMO it never will, so up to date analogue one is definitely superior in this respect.
.

Didn't you think why our genetic code appears to be grossly misdesigned from perspective of information theory?
It is inefficient, not binary - means uneconomic from data storage perspective, prone to countless translation errors, spontaneous and induced mutations etc.
However these are particular features which are allowing for species evolution and because of them life is creative.

So make an information processing system too perfect and it will merely deliver input - output processing, but show no ingenuity or problem solving ability.
That is a feature of digital systems.

You're missing the fact that the universe obeys predictable physical laws. These laws are expressible in mathematics and that means you can make arbitrarily accurate models in principle ab initio.You seem to be asserting that there is some magic creativity that can't be modeled in mathematics that only exists in the physical world, and we have no reason to expect thats the case. You're arguing from intuition.
User avatar
Dezakin
Heavy Crude
Heavy Crude
 
Posts: 1569
Joined: Wed 09 Feb 2005, 04:00:00
Top
Next
Post a reply

Return to Book/Media Reviews

Who is online

Users browsing this forum: No registered users and 1 guest



cron