FRIHOST FORUMS SEARCH FAQ TOS BLOGS COMPETITIONS
You are invited to Log in or Register a free Frihost Account!


AI





Newnone
can AI be created from decreased human nural pathways? i read somewhere in a fiction book that human pathways can be created to suit AI to 'think', whats yourr take in this?
_AVG_
Er, what exactly does AI mean?

And, if you read something in a fiction book, it's quite unlikely to be true.
AlmightySenator
I think that the problem with AI is that there is a "Bridge" that needs to be crossed. I've got degrees in computer science and philosophy (my college cancelled the cognitive science program), and looking from both ends, there is something that makes living beings alive. You can look at code, a language like Prolog or Lisp, and see that programs can "learn" with fuzzy logic and whatnot, but there isn't something there that makes them alive. If you've ever seen Blade Runner, you'll see that "something extra" that blurs the line between robot and human. I really don't think that it can get that far though, because there is something that is missing, and I don't think that humans have the ability to put it in.
KronikSindrome
I think we could get to the point where we could create an artificial intelligence....
but it will be just that...artificial.....and even then I think it would be far into the future
when bio-technologies are more advanced......

either way, when they make robots generally available and affordable....I'm gettin one!

I want a moon monkey too Laughing

http://www.rathergood.com/moon_song/
loryl
AI stands for Artificial Intelligence. I don't think creating AI from neural pathways will happen any time soon. Our brains are too complicated for mere machines!
Bikerman
Newnone wrote:
can AI be created from decreased human nural pathways? i read somewhere in a fiction book that human pathways can be created to suit AI to 'think', whats yourr take in this?


The question is a little confused. AI models use (amongst other things) neural networks; one theory in psychology holds that the brain is a sophisticated example itself of this type of technology. So in a sense existing AI systems may be considered as simplified models of the brain. The question of whether AI can be created depends on your definition of intelligence. In my understanding of the term, AI has already been with us for sometime. Systems already exist which display characteristics normally associated with intelligence - decision making, memory, behaviour modification in light of experience. The level of intelligence which such current systems display is, for sure, much less than that of an average human - some estimates would put the best of current systems at the intelligence level of an earthworm.
DoctorBeaver
I too have a degree in computer science with a PhD in psychology. My interest in psychology was aroused by studying AI as part of my computer science degree.

You don't need life for AI. Don't confuse consciousness with intelligence. And don't confuse robots with AI. Robots are just 1 possible application for AI.

I see no insurmountable obstacles with creating AI machines. The brain is pretty much mapped out and I'm confident that a major breakthrough in neuroscience that will allow us to figure out just how the brain processes information is not only attainable but not too far in the future - maybe even within the next 10-20 years.

Stem cell research could provide a method of growing a not-necessarily full-size brain, but 1 of sufficient capacity for limited cognition that could be linked to I/O devices.

As for the here-and-now, as has been stated by bikerman, systems already exist that display a limited amount of intelligence. One that springs to mind is in use on the London underground railway system to detect when platforms are getting crowded. CCTV images of the platform are fed into a neural network which analyses them and if it calculates that there are enough people on the platform it closes the entrance gates. The computer has learned to recognise overcrowding.

There has also been research at Brunel University just outside London with collision-avoidance robots that has produced unexpected results. The only instructions these robots are given are to keep moving and not to hit anything. Although the robots are identical, they develop different strategies for avoiding collisions. Some will reverse, others will turn left or right at varying angles. This and the London underground system are probably 2 of the best examples of AI in action.
AftershockVibe
I don't see any massive problem with there being working AI in the future either.

What most people think of when you say "AI" is a computer which mimics how a brain works. I feel the biggest problem with this at the moment is that computers were (up until recently) a single processor which is really fast and tailored to one purpose. ie - Processing is done almost entirely in series.

The brain however, can process what? thousands? millions? of things simultaneously so we're a long way off yet.
Bikerman
AftershockVibe wrote:
I don't see any massive problem with there being working AI in the future either.

What most people think of when you say "AI" is a computer which mimics how a brain works. I feel the biggest problem with this at the moment is that computers were (up until recently) a single processor which is really fast and tailored to one purpose. ie - Processing is done almost entirely in series.

The brain however, can process what? thousands? millions? of things simultaneously so we're a long way off yet.


There are other issues, though. Modern computer technology, interestingly, started out on the parallel computing path in the 80's and the first transputer chips soon followed. As it happens, however, the market was never sufficient to drive developments in that area very far and nowadays most general purpose computing has taken a slightly different track - using multiple ALUs within a single MPU/CPU. In other words the CPU chip now contains multiple Arithmetic/Logic Units and the instructions comprising the program are routed to different ALUs according to a set of rules. This type of 'parallelism' is correctly known as 'superscalar processing' and is the norm on machines other than supercomputers and clustered systems.
The brain also connects the different nodes in 3-dimensions rather than the 2-dimensional linkages used in transputer/superscalar technologies presently. That adds a massive amount of power since it introduces a whole new 'degree of freedom' to the signal paths.
It also, of course, reduces the signal-path-length which makes it an inherently faster and more efficient processing device since signal attenuation and lag/smearing are both minimised....

Such models are being developed but the massive complications make them expensive, fragile and fairly dedicated (ie not multi-purpose).
A whole new paradigm is necessary to describe the processing in such systems and, as yet, I don't think a standard has emerged.

Here is one attempt to standardise some of the concepts
Here's a website devoted to some of the issues
Here are some fun Java apps on the subject
newolder
Bikerman>
Quote:
"A whole new paradigm is necessary to describe the processing in such systems and, as yet, I don't think a standard has emerged."


The internet* is a working model of 3-d + time networks and the Grid will only improve matters. Has it (the Grid) switched-on yet? I didn't notice any flicker in the power supply... Smile

Another thing about these silicon-based networks (life-forms?) is that they approach brain-complexity with increasing accuracy as time goes by: the opposite of our disease-prone carbon based units. Mine is smashed most of the time these days but i enjoy the moments of lucidity too. Shocked

* ip6 address space is sufficient to give each individual an internet's worth... It's big...

http://www.circleid.com/posts/scarcity_ipv4_addresses/
Bikerman
newolder wrote:
The internet* is a working model of 3-d + time networks and the Grid will only improve matters. Has it (the Grid) switched-on yet? I didn't notice any flicker in the power supply... Smile
Well, hmm. Not quite. The internet is much more linear than a true 3-D net - it is largely a simple 2-D mesh topology with lots of hub-spoke connectivity. It is also very 'serial' in nature being almost entirely 'single-path' based because of the routing protocols currently used and most of the current grid implementations are done in middleware and still rely on the basic IP4 transport layer.
We are still a good way from true grid networking - even when IP6 is fully implemented (years) it will still be a good while before the rest of the infrastructure makes best use of the advantages it offers.
Most developers are only really looking at the increased address space offered by IP6 which is only really an extension of scale. The real benefits will come when the extra QOS 'hooks' are utilised and multicasting, combined with the ability to go point to point as well as the new routing improvements all begin to merge into true grids...
There's a basic introductory paper (PDF) HERE for the layperson who wants to know a bit more.
Quote:
Another thing about these silicon-based networks (life-forms?) is that they approach brain-complexity with increasing accuracy as time goes by: the opposite of our disease-prone carbon based units. Mine is smashed most of the time these days but i enjoy the moments of lucidity too. Shocked

Yep - my neural hardware will probably not last long enough to see the limits of the new possibilities that IP6 opens-up....Smile Still, judging by the use that a large part of the existing tech has been put to, it remains to be seen if the new topology will just become a super-media delivery system designed to replace the TV as the modern opium of the masses, or whether the real possibilities for social and political revolution will be realised.....
I can but hope Smile
wxzt
Do you mean the robots in the film "AI" ?
I don't think human can made it because humans are less clever than Nature.
DoctorBeaver
Maybe I'm missing something, but I don't see how the internet in any guise can enter into a discussion on AI. The net is merely a data storage/delivery system. It's quite possible, however, that some (maybe all) computers that are linked to the internet (or whatever the internet becomes in the future) could eventually be AI-based. But that still would not make the net an AI system.

I've got the StumbleUpon Firefox extension. On the surface it may appear intelligent as it delivers websites to my browser that it thinks I'd be interested in. However, the choices it makes on my behalf are based on my reaction to sites it presents me with; I say whether I like or dislike a particular site and it hones my category preferences accordingly. From the Turing machine point of view, it could be said that StumbleUpon displays intelligence; but, in truth, we all know that it works according to a pre-programmed set of rules. It doesn't actually "learn" my preferences.

In a previous incarnation I used to work on Expert Systems. Those also appear to make judgements whereas, in fact, they are no more than a series of rules governed by fuzzy logic and probability calculations. True, they did at times come up with surprising results; but the results could be analysed and the reasons for them established by examining the logic of the rules. There was no intelligence involved (other than mine for programming them in the first place! Laughing )
newolder
Bikerman>
Quote:
Well, hmm. Not quite. The internet is much more linear than a true 3-D net


Well, hmm. It's a question of definition then. Angel

As of now* t'internet occupies a volume in 3-space of the order 1 Solar system's worth.

It has a storage capacity undeamt in previous human endeavour**.

It has information anent all the best theory created by humans (and the worst).

It solved turbulence some time ago.

It's probably about to become self-aware.

It's built on maths and physics demonstrable to all.

Great fun, whilst it lasts :: http://dotsub.com/films/dancemonkeys/index.php?autostart=true&language_setting=en_404

ed.




* a system call that returns chronological data.

** Moore's law still applicable 20-odd years in.
EanofAthenasPrime
wxzt wrote:
Do you mean the robots in the film "AI" ?
I don't think human can made it because humans are less clever than Nature.


Nature isn't "clever" at all. Although I don't put much confidence in human intellegience, Nature has no intelligence at all. It was only through random chance that life even exists...
EanofAthenasPrime
Newnone wrote:
can AI be created from decreased human nural pathways? i read somewhere in a fiction book that human pathways can be created to suit AI to 'think', whats yourr take in this?


uh...wouldn't it be easier to synthesis organic matter instead of trying to compile half-decayed matter into a machine..?
Related topics
unbeliviable
Graphistes - J'ai besoin de vous!
Algum Brazuca ai?
DotA - Alguem ai joga?
un site sur des trucs que j'ai vendre
Help - SVG to AI or CDR Converter
Help - SVG to AI or CDR Converter
Seriado LOST
no consigo que aprovem meu lance
Codeur php/sql/js? j'ai besoin d'un coup de pouce
AI?
J'ai fais un nouveau site avec mes petits doigts maigrelets
J'ai perdu mon nom de domaine
Bon j'ai pas le temps de poster Ae Ae Ae
TNT
Reply to topic    Frihost Forum Index -> Science -> General Science

FRIHOST HOME | FAQ | TOS | ABOUT US | CONTACT US | SITE MAP
© 2005-2011 Frihost, forums powered by phpBB.