Artificial intelligence (Full Version)

All Forums >> [Casual Banter] >> Off the Grid



Message

KaineD -> Artificial intelligence (1/26/2009 2:28:34 PM)

I'm fascinated by the topic of artifical intelligence and whether or not it will one day be possible to create a machine that is aware of its own existence and aware of the world around it.  Consciousness in a machine.

Occasionally I'll surf the net for any information regarding the topic.  One interesting thing that people do is build computers to try and pass the turing test.  Basically what the turing test is, judges will use a chat program to chat with humans and computers, and if they are fooled and think the computer is a human, then the computer passes the test.  This used to be the standard for judging if a computer achieved AI, but now it seems the turing test is no longer adequate.  Give a program enough coding and phrases and responses, and it will fool a judge - but it doesn't mean the program isn't intelligent.

One thing I have come across though that looks promising, is a group of researchers that are trying to "reverse engineer" the human brain.  Sounds kinda kooky, but they're actually making good progress.  Using a program called BOINC, they have thousands of volunteers linked through the internet to lend the researches their collective computing power.  So what they're doing, is they've got a neural network simulator, and they've already been able to simulate over 100 billion neurons.

http://www.intelligencerealm.com/aisystem/system.php

So you can download BOINC if you want and lend these guys a little bit of computing power.  I think its pretty cool.  It seems like the most promising project for achieving artificial intelligence anyway.



JustDarkness -> RE: Artificial intelligence (1/26/2009 2:31:09 PM)

can we make a BOINC CM clan ;)

btw..is their a clients that uses nvidia cards through CUDA for this?
mmm yes it is the same I already run..for other projects

http://boinc.berkeley.edu/cuda.php

quote:

Two projects currently have GPU-enabled applications: GPUgrid.net and SETI@home. If your computer has a recent NVIDIA GPU, you can compute 2X to 10X faster than with your CPU alone



awmslave -> RE: Artificial intelligence (1/26/2009 3:52:49 PM)

quote:

group of researchers that are trying to "reverse engineer" the human brain

Do they know how brain works? If not, how do they reverse-engineer? To my knowledge nobody really knows how brain works.



philosophy -> RE: Artificial intelligence (1/26/2009 3:58:44 PM)

quote:

ORIGINAL: awmslave

quote:

group of researchers that are trying to "reverse engineer" the human brain

Do they know how brain works? If not, how do they reverse-engineer? To my knowledge nobody really knows how brain works.



Do we know precisely how the brain works? No.
Do we have a working knowledge of how large portin work? Yes. How do you think brain surgery or some pharmaceuticals work?

Of course we can reverse engineer those portions we understand, and we can learn from that process how other parts work better. Knowledge is not black and white, it is incremental...knowing how the brain works is one thing we have been working on for quite some time, so we have made progress.



awmslave -> RE: Artificial intelligence (1/26/2009 4:36:27 PM)

quote:

Do we have a working knowledge of how large portion work? Yes. How do you think brain surgery or some pharmaceuticals work?


To my knowledge brain surgery and pharmaceuticals are mostly based on trial and error method. Brain surgery just takes advantage of the known flexibility of brain (brain can restore function using left over parts). Narcotics have been around thousands years. As much as I know  we can measure brain activity and we have lot of knowledge about electrical and chemical impulses and neuron connections. I have not seen the model how brain stores and processes information. Maybe somebody has a good link to help me out. For example how exactly long- and short-term memories are stored, accessed and what type of algorithms are used?
To my understanding the "artificial intelligence" project you refer to is basically extended parallel processing project using existing computer hardware and principles well known.




samboct -> RE: Artificial intelligence (1/26/2009 4:43:08 PM)

I dunno about artificial intelligence, but we've been very successful at creating artificial stupidity.  I don't know why, the natural stuff isn't in short supply....



philosophy -> RE: Artificial intelligence (1/26/2009 7:55:34 PM)

..for your edification....

http://en.wikipedia.org/wiki/Human_brain

.....a certain amount of what we know about the brain came from trial and error, but many things have now been worked out. i seriously doubt modern brain surgeons could get away with advocating brain surgery so they can stick a scalpel in and wiggle it about to see what happens. You  may have read too many Glen Larson cartoons, which i thought impossible.



Vendaval -> RE: Artificial intelligence (1/26/2009 10:13:01 PM)

If AI does not exist then why do I waste so much time sweet-talking old machinery and cursing at tech support?  



corysub -> RE: Artificial intelligence (1/27/2009 6:20:52 AM)

I don't think you will ever be able to create a program that does anymore than solve complex problems.  We can put many things on automatic pilot, have robots that can build an increasing percentage of a car structure, fill bottles, and turn on the coffee in the morning.  Those are things we could figure out with a slide ruler if given enough time...but a computer can never be taught to "think creatively", out of the box, and invent, in my opinon. 



kittinSol -> RE: Artificial intelligence (1/27/2009 6:27:28 AM)

quote:

ORIGINAL: corysub
a computer can never be taught to "think creatively", out of the box, and invent, in my opinon. 


You'd be surprised. The time of the machines will come.



CatdeMedici -> RE: Artificial intelligence (1/27/2009 6:34:42 AM)

Artificial intelligence is designed and programmed by humans and will always fall short of perfection--it will only be as good as the best human working on it.  IMHO.



KaineD -> RE: Artificial intelligence (1/27/2009 6:46:05 AM)

quote:

ORIGINAL: CatdeMedici

Artificial intelligence is designed and programmed by humans and will always fall short of perfection--it will only be as good as the best human working on it.  IMHO.


Not if it has the capacity to learn and grow.



FRSguy -> RE: Artificial intelligence (1/27/2009 11:35:43 AM)

Use to work on A.I.. as a kid.  Its really pretty easy for the most part especially with text conversing.  Its vision and audio that can be a real bitch. As far as a machine that is self aware thats been around since the 60's and is very easy to do.  I did it when I was kid by creating two programs... when the A.I. program searched for an enviromental answer the second program would kick in and invent an answer for it.  So lets say you asked it... what does the wallpaper in your living room look like?  The second program would kick in and add the detail to a simulated enviroment and log it in.... although its fake the A.I. would believe that things were allways a certain way.  All the programs I wrote would melt down at some point however... but mood swings ect. was pretty easy by using bio-rythem software against a sentence scoring system that would score sentences for emotional impact. Its only when you try to get your software to truly look and listen to the outside world that things get difficult.



philosophy -> RE: Artificial intelligence (1/27/2009 11:54:40 AM)

quote:

ORIGINAL: kittinSol

quote:

ORIGINAL: corysub
a computer can never be taught to "think creatively", out of the box, and invent, in my opinon. 


You'd be surprised. The time of the machines will come.


...current computer technology is, as we all know, based on a binary paradigm. The circuit is either on or off, and we have become adept at creating mechanisms where '0's and '1's become more complex things. However, if we ever get quantum computing to work, we'll be switching to (if my memory serves) 32 possible states. The resulting complexity and its attendent rise in the possibility of emergent behaviour makes AI a much more likely prospect.

...heres an article on the subject which i can barely understand....lol
http://en.wikipedia.org/wiki/Quantum_computer



DomKen -> RE: Artificial intelligence (1/27/2009 12:07:35 PM)

quote:

ORIGINAL: corysub

I don't think you will ever be able to create a program that does anymore than solve complex problems.  We can put many things on automatic pilot, have robots that can build an increasing percentage of a car structure, fill bottles, and turn on the coffee in the morning.  Those are things we could figure out with a slide ruler if given enough time...but a computer can never be taught to "think creatively", out of the box, and invent, in my opinon.

This is simply wrong. AI software already invents.

Investigate genetic algorithms 



DomKen -> RE: Artificial intelligence (1/27/2009 12:18:29 PM)

quote:

ORIGINAL: philosophy


...current computer technology is, as we all know, based on a binary paradigm. The circuit is either on or off, and we have become adept at creating mechanisms where '0's and '1's become more complex things. However, if we ever get quantum computing to work, we'll be switching to (if my memory serves) 32 possible states. The resulting complexity and its attendent rise in the possibility of emergent behaviour makes AI a much more likely prospect.

...heres an article on the subject which i can barely understand....lol
http://en.wikipedia.org/wiki/Quantum_computer

The advantage in quantum computing isn't in the range of numeric value of each stored bit, if a 0 to 31 value is desired a computer could easily be designed to operate on 5 bit groups which quit eobviously allows for the same 32 values. Quantum computings theoretical advantages lay in the fact that a group of quantum bits can be in a number of different states, values, simultaneously. There are additional ramifications of using quantum states as storage that I don't fully understand.



awmslave -> RE: Artificial intelligence (1/27/2009 1:20:48 PM)

quote:

i seriously doubt modern brain surgeons could get away with advocating brain surgery so they can stick a scalpel in and wiggle it about to see what happens. You may have read too many Glen Larson cartoons, which i thought impossible.


Believe it or not but this is what they do. A patient is awake during the surgery and the communication with a patient tells a surgeon how to proceed.



philosophy -> RE: Artificial intelligence (1/27/2009 1:20:48 PM)

quote:

ORIGINAL: DomKen

There are additional ramifications of using quantum states as storage that I don't fully understand.


...glad it's not just me then. lol. However, i think the main point is that this new technology promises to be able to do things that the old one couldn't. AI is probably one of them.



philosophy -> RE: Artificial intelligence (1/27/2009 1:22:07 PM)

quote:

ORIGINAL: awmslave

quote:

i seriously doubt modern brain surgeons could get away with advocating brain surgery so they can stick a scalpel in and wiggle it about to see what happens. You may have read too many Glen Larson cartoons, which i thought impossible.


Believe it or not but this is what they do. A patient is awake during the surgery and the communication with a patient tells a surgeon how to proceed.



...only for some procedures. Not for all of them. And even then, the surgeon does know the general area to fiddle with. So, your thesis that we know nada about the brain is clearly claptrap.



awmslave -> RE: Artificial intelligence (1/27/2009 1:34:41 PM)

quote:

So, your thesis that we know nada about the brain is clearly claptrap.


It is not what I think. We know a lot. I just am unable to find much information how brain actually processes  information. Wiki article tells nothing about it. How and where short and long term memories are stored are accessed for example?



Page: [1] 2   next >   >>

Valid CSS!




Collarchat.com © 2025
Terms of Service Privacy Policy Spam Policy
0.046875