Subject:
New Way to develop AI
Author:
anonymous
Date:
2/5/2004
AI ( Artificial Inteligence )
2/2/04
by Kane Jun
*** Background ***
My hope to writing this is share my idea with others. This is only one form of AI. I copied many idea from our own human intelligence. I reasoned that if human brain function not randomly but orderly ( i.e. according to some rule if not mathmetic or logic ) that rule can be thought out. If it is random then there would be no human inteligence. The whole new branch of science have to created. It consists of manipulation of concept.
My experiences tell me that the human intelligence is 99% visual. When we read books we comprehend by visualizing the situation. Since currently we can not store visual intelligent data in the computer we are going to have to charactorize the visual experience(i.e. the visual data have to be written). So the data is string of charactors which represent visual data.
*** CONCEPT ***
The concept is the term I prefer to use since there is no word that can explain exact meaning of it(i.e. it is closest). The concept is the basic unit of information or data. For human intellingence the concept is instance of thought when input comes in throgh our sensory system. When it comes in it usually comes in with network of information ( or association). This includes feeling but I am going to omit it(Reasoning intelligence is good start to make AI). Since parallel input is not possible currently we can simulate it with multiple of serial input. In the computer program we can use string of charactors as concept. I use the terms concept and string exchangably.
The example of string of charactors are letter or document written using word processor. The whole idea of my AI is to manipulate (i.e. add function to ) these string. We should have very big collection of string which represent memory which can be saved opened using program. After the manipulation the whole string of charactors change from one state to another just as our memory do. In the future if we have very large storage and very high speed we just need good AI software.
*** Function and Value ***
At the lowest level there are very basic function and value.
The very basic function are SHORT FORMING and MATCHING. The derived functions are the functions that are derived from the basic functions(short forming, matching and other basic functions). The examples of derived function are LEARNING, GENERALIZATION, PATTERN MATCHING and NEW IDEA CONSTRUCTION. These derived functions are sometimes made by combining basic function and basic values together. The derived functions can be combined to make up even higher derived function.
Few examples of very basic value are Good feeling, Bad feeling. Since the computer does not have any feeling we can simulate them by just words GOOD and BAD. There are other values such as Infinity etc...
Basic function :
RETRIVAL : The retrival is simply recalling the memory.
TEMPERARY STORAGE : This is like our (human) short-term memory.
MATCHING : The m
|