View Full Version : Artificial brain 'ten years away'
Morbus
07-23-2009, 01:30 PM
http://news.bbc.co.uk/2/hi/technology/8164060.stm
This is from Henry Markram, whose Blue Brain project successfully copied the functionality of a single neocortical column (of a rat) and is now moving on to a whole human brain
It's quite a leap, but since the neocortex is pretty similar across mammals, and since the neocortical columns are pretty similar across any given brain, believe it or not, they probably got a lot of the hard stuff out of the way already
artificial brain probably. Intelligence, though? keep dreaming
stevensonmat2
07-23-2009, 01:35 PM
Yeah, I mean only god can make intelligence.
if someone manages to program a working intelligence in 10 years I'll be extremely surprised. Especially considering how shitty all the programming being done nowadays is.
Morbus
07-23-2009, 01:47 PM
if someone manages to program a working intelligence in 10 years I'll be extremely surprised. Especially considering how poopty all the programming being done nowadays is.
Uhhhhhhh
I hope you realize that there is a heavens-wide difference between the explicit, ad hoc programmatic approach to artificial intelligence and what Markram et al. are doing in Switzerland
bottom-up has potential, but not in 10 years
top-down is a lost cause
Morbus
07-23-2009, 01:50 PM
bottom-up has potential, but not in 10 years
Well they delivered the first stage on time
Do you know how similar human neocortical columns really are?
quite similar, but still, even if they build a brain it doesn't necessarily mean they will have a working intelligence anytime soon.
Morbus
07-23-2009, 02:01 PM
quite similar, but still, even if they build a brain it doesn't necessarily mean they will have a working intelligence anytime soon.
......
ok, three minutes have passed and all I can come up with is: huh?
Well, probably he means that brain is just a framework in which intelligence is capable of working at. It is a very silly assumption that artificial brain = artificial intelligence, since it is evident that even human brains in certain circumstances can't reach the level of intelligence of a human (wolf childs etc.).
When we have the artificial brain (or any other similiar concept), the correct training algorithms and means of communication we might one day come up with something which seems intelligent.
Well, probably he means that brain is just a framework in which intelligence is capable of working at. It is a very silly assumption that artificial brain = artificial intelligence, since it is evident that even human brains in certain circumstances can't reach the level of intelligence of a human (wolf childs etc.).
When we have the artificial brain (or any other similiar concept), the correct training algorithms and means of communication we might one day come up with something which seems intelligent.
this is what i meant. I'm a bit groggy still so I wasn't able to put everything into words.
Berner
07-23-2009, 02:30 PM
Well, probably he means that brain is just a framework in which intelligence is capable of working at. It is a very silly assumption that artificial brain = artificial intelligence, since it is evident that even human brains in certain circumstances can't reach the level of intelligence of a human (wolf childs etc.).
When we have the artificial brain (or any other similiar concept), the correct training algorithms and means of communication we might one day come up with something which seems intelligent.
It's a hell of a start though.
Sure, and after reading the article I do agree that artificial brain is good gadget for finding cures for neurological diseases.
Morbus
07-23-2009, 02:35 PM
Well, probably he means that brain is just a framework in which intelligence is capable of working at. It is a very silly assumption that artificial brain = artificial intelligence, since it is evident that even human brains in certain circumstances can't reach the level of intelligence of a human (wolf childs etc.).
When we have the artificial brain (or any other similiar concept), the correct training algorithms and means of communication we might one day come up with something which seems intelligent.
s/seems/is/
Based on what we already know about, e.g., neurons cultivated on an electrode array to serve as a computer, Vernon Mountcastle's "An Organizing Principle for Cerebral Function," experiments with swapping visual and auditory cortices in rats, etc. it seems neocortical tissue just kind of automatically responds to whatever you give it. It learns by itself.
So, come 2020, we just need to start throwing various sensory inputs at the final instantiation of Blue Brain and it will make sense of them, just like we do.
And that exactly is the problem, intelligence is so much more than sensory input and output. Neural networks are good at managing sensory data just because the mathematics are very very similiar. For intelligence we need so much more.
Morbus
07-23-2009, 02:52 PM
And that exactly is the problem, intelligence is so much more than sensory input and output. Neural networks are good at managing sensory data just because the mathematics are very very similiar. For intelligence we need so much more.
The thing is that the algorithm used throughout the neocortex doesn't change too much. I mean it does change from one area to the next, but not by a lot.
If you crack open someone's skull, you'll notice that the different regions of the neocortex don't look very different
And for the most part they really aren't
At heart, cognition (which is incestuously tied into the senses btw) isn't all that different from sensory perception
Cognitive science is still at the stage where, like, science as a whole was, in the day of Aristotle.
It's gonna be a hell of a lot longer than 10 years.
Morbus
07-23-2009, 04:15 PM
Cognitive science is still at the stage where, like, science as a whole was, in the day of Aristotle.
I don't think you can seriously equate the state of the art in neuroscience (which isn't the same thing as cognitive science by the way: confusing the two already undermined the credibility of your post) to Aristotle saying "heavier objects fall faster than lighter objects"
Mmmmmmm ... it's a bit more than that
You also have to consider the tempo of modern neuroscience is significantly faster than anything in Aristotle's day
I don't think you can seriously equate the state of the art in neuroscience (which isn't the same thing as cognitive science by the way: confusing the two already undermined the credibility of your post)
They're intimately connected
to Aristotle saying "heavier objects fall faster than lighter objects"
The point is, we're barely scratching the surface of what there ultimately is to discover
Morbus
07-23-2009, 04:33 PM
Yeah but you haven't offered anything against Markram's timeframe besides "nuh uh no way not gonna happen"
That's a rebuttal by fiat. It's not convincing
I mean he was on time for phase one ... this is in Switzerland after all, you bet your bottom dollar he'll be punctual
lukeskywalkertakingadump3
07-23-2009, 06:16 PM
Not interested, I want a Gundam.
suspect device
07-23-2009, 06:53 PM
the word artificial sums it up, thats what it will always be
Yeah but you haven't offered anything against Markram's timeframe besides "nuh uh no way not gonna happen"
That's a rebuttal by fiat. It's not convincing
I mean he was on time for phase one ... this is in Switzerland after all, you bet your bottom dollar he'll be punctual
Well there's no way I can prove it's not going to happen
But the fact that we are barely scratching the surface of how the brain functions and how the mind is structured strongly suggests that it will not happen in 10 years
The thing is that the algorithm used throughout the neocortex doesn't change too much. I mean it does change from one area to the next, but not by a lot.
If you crack open someone's skull, you'll notice that the different regions of the neocortex don't look very different
And for the most part they really aren't
At heart, cognition (which is incestuously tied into the senses btw) isn't all that different from sensory perception
Yes, I do agree that the level of intelligence is related to the capability to perceive our environments. But mimicing the neocortex is no where near artificial intelligence.
Smokey D
07-24-2009, 12:03 AM
From my reading of the article, it suggests that they're going to have a functionalist model of a human brain, not a brain which is in fact equivalent to a human brain.
Against Miik!
07-24-2009, 12:10 AM
You guys I just spent like 3 days ridding my computer of a virus so if we are ten years away from making a freakin brain I'd be pretty surprised.
Also, there will probably be a republican president in the WH in the next 10 years so js
Morbus
07-24-2009, 05:18 AM
the word artificial sums it up, thats what it will always be
artificial, adj.
not arising from natural growth or characterized by vital processes
Hohoho, you got me there!
And your point is?
Well there's no way I can prove it's not going to happen
But the fact that we are barely scratching the surface of how the brain functions and how the mind is structured strongly suggests that it will not happen in 10 years
The only people I'll trust for timeframes are those working on the project themselves, since they can probably give the most realistic assessment of their own research.
Yes, I do agree that the level of intelligence is related to the capability to perceive our environments. But mimicing the neocortex is no where near artificial intelligence.
And do not think you can fool me in this case, I am developing neural networks algorithms for determining social music profiles of bands as my profession.
Yeah ok that's cool but Blue Brain isn't a giant backpropagation model durr hurr js
From my reading of the article, it suggests that they're going to have a functionalist model of a human brain, not a brain which is in fact equivalent to a human brain.
Functionally equivalent
Same thing
You guys I just spent like 3 days ridding my computer of a virus so if we are ten years away from making a freakin brain I'd be pretty surprised.
Also, there will probably be a republican president in the WH in the next 10 years so js
This isn't the same thing as Windows XP
And it's based in Switzerland
artificial, adj.
The only people I'll trust for timeframes are those working on the project themselves, since they can probably give the most realistic assessment of their own research.
Or overblown because they have egos and too much ambition and want funding
Morbus
07-24-2009, 02:25 PM
Yeah I'll bet they spend that money on Porsches, caviar, coke and 3000 dollar a night hookers too
I've noticed that people who deride progress have two modes:
If it hasn't been achieved yet: criticize
If it has: backpedal
This is so productive, guys!
free_thinkers_are_dangerous
07-24-2009, 02:43 PM
10 years would be shocking, but anybody who's trying to diss this achievement based on how fast 10 years sounds is dumb. It sounds fast, but who knows?
Morbus
07-24-2009, 03:18 PM
Compare 1980 to 2000
Quite a difference huh?
Why should we think 2020 will not be substantially different from 2000?
No let's just pretend that nothing radically new will ever happen, it's comforting, man
Light Flantastic
07-24-2009, 03:21 PM
there will probably be a far greater difference between 2000-2020 than 1980-2000
as long as they make roomba less retarded we cool
Well, actually their approach seems interesting, hopefully there is enough neurological data to continue the project towards human brain simulation.
However, saying that neocortext is very similiar between any mammal is a bit bold statement, if we want to include humans. Also the mathematics for sensory perception, motor commands and spatial reasoning is easy and for conscious thought and language are very much do able. However, if they want to use their project for research on mental health medication I believe some forms of cognition are required, which might not be as easy because cognition - which probably is essential component for such - requires much more from associative memory and complex classification of "experience".
Morbus
07-26-2009, 08:06 AM
However, saying that neocortext is very similiar between any mammal is a bit bold statement, if we want to include humans
Bold but fairly true
At least if we don't include the cognitive side of things. Humans are very similiar in the ways we percept the visual and aural world and how we use our muscles.
Morbus
07-26-2009, 08:23 AM
Anatomically speaking, afaik, neocortical modules are pretty similar across all mammals
suspect device
07-26-2009, 05:00 PM
artificial, adj.
not arising from natural growth or characterized by vital processes
Hohoho, you got me there!
And your point is?
well i think that they may very well create processes that reproduce an artifical brain function. same inputs, same outputs but the outputs will be predefined by the programmers
the clearest way that machines will in anyway take some sort of control of human destiny is if we get lazy and hand it off to them. otherwise they will always be dependent on humans for instruction
10 years would be shocking, but anybody who's trying to diss this achievement based on how fast 10 years sounds is dumb. It sounds fast, but who knows?
You guys are taking it like a personal attack that I'm saying it's probably not going to happen Christ
we'll find out in 10 years anyway, won't we?
Smokey D
07-27-2009, 06:34 PM
Functionally equivalent
Same thing
Well, as I say, on my reading of the article, that's not what they're saying. A computer which can model certain features of certain mental illnesses is not the same as intentionality.
Chris Is Gay
07-30-2009, 02:31 AM
well i think that they may very well create processes that reproduce an artifical brain function. same inputs, same outputs but the outputs will be predefined by the programmers
:lol:
Ok you don't understand any of this, I'm done with you
Chris Is Gay
07-30-2009, 02:32 AM
Well, as I say, on my reading of the article, that's not what they're saying. A computer which can model certain features of certain mental illnesses is not the same as intentionality.
Listen to the dialogue with Markram at the end
TerranYouApart
07-30-2009, 02:33 AM
is masturbating with a cum sock the same as having a sex?
(i wouldn't expect you to know the answer to this)
an artificial brain is exactly that. artificial
Chris Is Gay
07-30-2009, 02:48 AM
Oh good job chad, false analogy
http://www.fallacyfiles.org/wanalogy.html
What, O Wise Neuroscientist Already_Taken, would distinguish an artificial brain from an evolved one
TerranYouApart
07-30-2009, 03:01 AM
maybe the same thing that distinguishes orange fanta from tropicana?
Chris Is Gay
07-30-2009, 03:05 AM
Oh great another weak analogy
Only you and chad could use outlandish analogies like these
What is that "thing"
TerranYouApart
07-30-2009, 03:12 AM
the fact that it is artificial is what separates it i thought that was so blatantly obvious that i used another "dumb" analogy in hopes you could pick up on my irony.
Chris Is Gay
07-30-2009, 03:15 AM
the fact that it is artificial is what separates it i thought that was so blatantly obvious that i used another "dumb" analogy in hopes you could pick up on my irony.
So what if it's artificial, doesn't mean it can't have thoughts and feelings
stevensonmat2
07-30-2009, 07:31 AM
Terran I have to say you are missing a few planks on the deck.
free_thinkers_are_dangerous
07-30-2009, 05:51 PM
the fact that it is artificial is what separates it i thought that was so blatantly obvious that i used another "dumb" analogy in hopes you could pick up on my irony.
Q: why is something artificial different from something that is natural?
A: because it is artificial
if this were grade 7 history you would fail. justify or stop wasting people's time and bandwidth.
TerranYouApart
07-30-2009, 06:05 PM
well i mean that's just the obvious answer. nevermind the fact that we have really only scratched the surface on why the brain works the way it does which i think is necessary if you want to engineer your own brain for the purpose of......
plus there are so many intangible things that go through any persons brain that i doubt it is possible to perfectly replicate a human brain from scratch.
if you wanted me to explain deeper all you had to do was ask.
Led_Zep_Bonham
07-30-2009, 07:57 PM
Another step towards the apocalypse, party time.
plus there are so many intangible things that go through any persons brain that i doubt it is possible to perfectly replicate a human brain from scratch.
Intangible doesn't mean non-physical
TerranYouApart
07-30-2009, 11:47 PM
a sub-conscience?
how does one even look into their sub-conscience to know how to apply it anywhere else?
Chris Is Gay
07-31-2009, 10:08 AM
plus there are so many intangible things that go through any persons brain that i doubt it is possible to perfectly replicate a human brain from scratch.
a sub-conscience?
how does one even look into their sub-conscience to know how to apply it anywhere else?
Even the shitty neural networks we have today can end up with different weights during training and yet converge to the same input-output mapping. Although we are unaware of some of the nitty-gritty details of this process, we know the gist of how it works, enough to replicate it in countless commercial applications every day. That's kind of an analogy to being able to use "intangible" or "subconscious" mental processes.
As for raising distinctions like conscious/subconscious, thinking/emoting, that's like creationists admitting that "microevolution" happens but "macroevolution" doesn't. These distinctions are meaningless. In the same way that macroevolution is the same thing as microevolution carried out to speciation, playing chess and getting pissed / gloating over the outcome of the game are different kinds of electrical activity in the brain.
And here another analogy to creationists is relevant. Why do people raise these objections even after having been presented with the facts, or make half-assed concessions like "an artificial brain could be great at math, but not have feelings"? Well, like creationists, this kind of behavior owes partly to persistent ignorance, but more so to wanting to belieeeeeeeve.
vBulletin® v3.8.1, Copyright ©2000-2009, Jelsoft Enterprises Ltd.