Artificial Intelligence

  Guestbook Androids Turing Components Links/Resources

Artificial Intelligence is:

The ability of an object to start or change actions from input information on an ongoing event.

The science of making machines do things that would require intelligence if done by people. [1]

Is based on knowledge and information.

One main problem is creating good representation of the knowledge to be processed.

Information can be represented by:

  1) links and associations  

 2)     Rules which relate to the above
 3)     Strategies for applying the rules  

                                                             

Semantic Networks show sets of relationships between objects  

Production rules use condition then action sequences i.e.

^ if < condition > then <action > e.g. if <not drilled> then <drill>  
A lot of development in AI has been made in game play such as chess. Game programs are based on searching for moves and countermoves before selecting the best.  In games such as noughts and crosses all moves can be searched and analyzed but in games such as chess this isn’t possible so a balance must be struck between times spent searching and analysing before making a move.  
Reasoning programs solve problems to do with pattern recognition and logic problems.

 
Some computer programs recognise limited voice input and process the information on a word processor; these systems have cumulative knowledge so that later corrections are saved and the system becomes more accurate.  

 
 The Four Colour Theorem, which says that any map can be shaded with just four colours so that no colour touches an area of the same shade was proved in 1967 by computer.  

An expert system is able to draw a reasoned conclusion from knowledge in a particular field and also communicate to the user how it reached its answer. This could be used for things such as medical diagnosis

  

To the top...

[1] Computing Science 2nd Edition by Peter Bishop