Artificial Intelligence

Recursive Search - How AI (artificial intelligence) Works To Calculate Load-Points & Stresses In Stretched-Domestic & Multi-Level Buildings & Structures

Up until now computers themselves have used a technique called recursive search in order to perform complex tasks such as playing a game of chess. Most high level programming languages support recursive function calling. This means that they use loops to call functions to test an infinte number of variables and keep going through the loop until a finite answer is reached. In a situation such as playing chess, the computer must go through all of the possible moves and outcomes until it finds the best option for a given situation. In 1997, this method was proven successful when the computer Deep Blue beat world chess champion Gary Kasparov. This was the first time ever that a computer had beaten the world champion in chess.

Neural Net Computing

Recursive search is good for some functions but it can become longwinded and therefore counter-productive. If we compare that with how a human being would think about things, there is a significant difference.

This is because of the way in which the human brain functions. We have a millions of neural pathways and we learn by effectively building bridges between them.
Imagine a situaton where someone throws a child a ball. He or she watches the ball but runs in the opposite direction and misses. When this happens, a neuron fires an electrical impluse which tells them that this is the wrong direction to go. The next time the ball is thrown, a neural pathway has now been built to tell the child that running the opposite way will not work. And so, the process continues until the child has caught the ball and the correct bridge has been built. That is how a human being learns as a child to catch a ball. The term "neural net" refers to the complex array of pathways and connections which fully grown adults has developed when they reach maturity.

Extensive studies have been made on the human brain and neural mapping has commenced. It is estimated that by the year 2020, we will have a complete map of the human brain. Imagine what we could do if this map could be transferred to computing. This has potentially mind boggling consequences because it means that this combined with raw computer and recursive search, could create a new type of machine intelligence or intelligent machine.

These "neural net" style systems are already being tested and in fact, this has been taken one step further with a method called "Organic Computing". Computers are acutally being developed in special labs which use actual protein-based tissues to send electrical impulses to replicate control signals in place of transistors. This is alo very exciting as it means that were these systems to become mainstream, they would be much more robust and would fucntion with out any of the temperature and environment restricions which inhibit transistors and silicone. Recursive search, neural nets and also quantum computing combined will be a powerful force in bringing human beings and computers ever closer.