Question for our resident computer scienctists in the field of A.I

Cpt_pineapple
atheist
Posts: 5492
Joined: 2007-04-12
User is offlineOffline
Question for our resident computer scienctists in the field of A.I

We know the brain uses electrical signals to do it's job. However, my question is can we replicate it? Can we make a robot that has emotions as we do? Can a robot feel, think, etc...

 

In other words, can we make a conscious A.I?


Magus
High Level DonorModerator
Magus's picture
Posts: 592
Joined: 2007-04-11
User is offlineOffline
 My AI programming has

 My AI programming has been limited to what a video game should handle.  Black Vs White, has some learning algorithms in it.  Perceptons are the most elaborate coding I have done as far as AI. Also know as the a learning algorithm.  I have just started in the "industry", so I might just be behind the times of what is useful in games.

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

Sounds made up...
Agnostic Atheist
No, I am not angry at your imaginary friends or enemies.


qbg
Posts: 298
Joined: 2006-11-22
User is offlineOffline
Note: I am far from an

Note: I am far from an expert.

Can we replicate it? Well, if all the brain is just one big neural network, then maybe (right now it would very very very far from real time assuming you have enough resources for the simulation).

Robot with emotions? We can try to simulate emotions (I recall a robot that was designed to do that).

Conscious? Well, what is consciousness and how does consciousness work are two questions that to my knowledge have not been answered. I know there is the Turing Test (no AI has passed it yet), but how does that relate to consciousness?

"What right have you to condemn a murderer if you assume him necessary to "God's plan"? What logic can command the return of stolen property, or the branding of a thief, if the Almighty decreed it?"
-- The Economic Tendency of Freethought


Cpt_pineapple
atheist
Posts: 5492
Joined: 2007-04-12
User is offlineOffline
But, how far can these

But, how far can these learning algorithms go? How much can the A.I actually learn?


stillmatic
stillmatic's picture
Posts: 288
Joined: 2007-03-29
User is offlineOffline
I'm not an expert in the

I'm not an expert in the field of AI either, but there are many people working on it like Jeff Hawkins.

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

 He's the man responsible for Palm Pilots and Treo's. He has also founded another company called Numenta that is developing pattern recognition software modeled after cortical layers in the human brain.

 I'm not a neuroscientist either, but I'm pretty sure that the functionality of the human brain isn't exactly entirely mapped out or understood. It would be quite difficult to create a model of the human brain in a computer without having an accurate model of the human brain.

Maybe I'm talking straight out of my ass about the human brain, my field is on the computer side.

"A proof is a proof. What kind of a proof? It's a proof. A proof is a proof. And when you have a good proof, it's because it's proven." -- former Canadian Prime Minister Jean Chretien


inspectormustard
atheist
inspectormustard's picture
Posts: 537
Joined: 2006-11-21
User is offlineOffline
Cpt_pineapple wrote: But,

Cpt_pineapple wrote:
But, how far can these learning algorithms go? How much can the A.I actually learn?

If you really want to know more, check libraries or Amazon for Artificial Intelligence: A Modern Approach by Stuart Russel and Peter Norvig. That was the textbook for our introductory AI class. It has an aweful lot of equations, pseudocode, and things like that but overall it's not so intense that anyone without a background in computer programming couldn't understand. I hope.

 

Barring that I'll try to explain as best I can. The limit is not in the algorithm but in the hardware. Even a relatively complete evolving algorithmic system can eventually replicate intelligence. However, given our current computer speeds and memory limitations it would be very very slow. It is evolutionary algorithms that are used to invent things automatically, and could be compared to our own creative areas in a way.

Next would be expert systems, which aren't really all that intelligent. They're just a huge collection of facts combined with a basic logic engine which is able to make comparisons like "if A is a square, then A is not a circle." Programs like ELIZA and our own emotions would fall into this category.

The fastest system we know of so far is artificial neural networks. Neural networks are capable of performing everything the above methods do, as they form a sort of self connecting/adjusting circuitry. It is possible to build hardware capable of acting as an ANN. We've been using radio waves in place of axons, the connecting fibers between neurons which are dynamic. You can see part of the formulae in my picture.

Once we're done with the prototype you may be hearing about it in the news. That is, as long as money holds out.