Follow by Email

Tuesday, May 15, 2007

A computer can't think, because of this

What usually human's think about ? talk about ? dream about ? live for ?
I've been thinking for several years about the process of human learning and perception. I gained several thoughts about it.
Regardless of humor, compassion, irony, and recalling memories; the rest of human brain power is directed towards what they don't know, being the most famous factor: the future.
Tomorrow's weather, the match result, the exam result, the future job, the future salary, the future inventions, can we someday reach the stars ?
And past mysteries: who murdered President Regan, El-Sadat, etc. How did the pharaohs build the pyramids.
Even gossip talks: Who said what about who, who cheated in the exam etc.
In the exception of the last point, the first 2 are usually lead science forward.
You would NEVER - almost ? - find someone who is interested in hearing ( or thinking ) what he already knows, stating the obvious they call it.
Who would care to know "again" that the High Dam gives Egypt electricity ?
But they care to think "what if it was destroyed?"
Humans usually think in stuff they don't know. And as paradoxical as it might seem, they need to know what they don't know so they can think about it !
So in my point of view, a computer system, to be able to think, and represent knowledge, it can not ignore the need to represent ignorance !
I might put it simpler: knowledge representation won't be perfect unless it can represent what is not known !
That's already the case in QAS ( Question Answering Systems ) when the represent questions. But can a computer system, generate such questions ?
It did, by comparing the input knowledge representation to a pre-defined cases of full meaning representation patterns, so it only meant to complete the given pattern.
But can the computer think beyond that ?
Till now that is not thinking. I mean can the computer system generate the full-meaning representation patterns by itself, and be able to induce the answer to the generated question by using common expert system techniques, and perhaps mathematical induction rules ?
Can it afterwards confirm the results of its own induction by launching virtual simulations and comparisons to basic scientific rules and maybe - sometimes - common sense ? ( I might write another one about common sense isA )
"generate the full-meaning representation patterns"... based on what ? Creativity. Can computers be creative ?
"generate the full-meaning representation patterns"... what will trigger the generation anyway ? What guides the direction of generation ? Might we say genetic-algorithms ? Is randomness the key ? I think this is strange, it might, and might not be the direction, but I don't feel like it is the correct way.


MeshMesh said...

Nice post....Actually, as we all know a computer will never reach the level of human thinking no matter what coz it doesn't just involve algorithms and experts programming it in the most clever way possible ..Yeah, u may argue that it has incredible computation speed,power and the like from the "body" side ...but our brain is a function of our whole body with its different systems,and yeah they realized that by improving computers through Artificial Intelligence ...specifically "Neural Networks" bt to understand the whole functioning of the body and how every part of it relates to our brain and all that knowledge seems enormously huge and suppose, they managed to reach such far-sought knowledge, what would be the final shape of computers and what abt their inner functionality ..what inner "chemicals" will it then be composed of?

We are all a part of our Creator's (God Almighty) greatness, creativity and we are full of wonders we'll never know abt us (like our souls)...but NO matter what far a human can be perfect inside out , he'll never be any degree close to His Creator's greatness and perfection

Yeah its not at all the least bit of comparison but look at the concept...Computers are a part of us (its "creators")..and so No matter how far a computer will come to be almost great from the "thinking" viewpoint it'll never be close to the least clever person on Earth so how will it be compared to the more intelligent ones?

But these are all some brief arguments...the topic does require more knowledge, more details and more and more subtopics ..I think I may write about it in my blog when I get to it deeper after exams i.s.Allah as this is a very interesting topic indeed

Jazak Allah Khayran...Pls keep it up!

Mohammad Nabil said...

I also have thought in a generalization of this theory, I named it "The Meta-ization paradox". I might write about it sometime.