Approaches to AI
There are two main different approaches to creating AI. These theories both justify their theoretical and practical use. These theories are the bottom-up method and the top-down method.
- Researchers working with the bottom-up method believe the best way to create artificial intelligence is to build electronic replicas of the human brain's complex network of neurons.
Researchers working on top-down approach are attempting to make a duplicate of the human brain's behaviour with computer programs.
The Bottom-Up Approach will be explained first:
The human brain is a web of billions of neurons. This network of neurons is what gives humans intelligent characteristics. Researchers who follow the bottom-up method of creating AI build electronic circuits that act as neurons do in the human brain. A lone neuron is useless but when grouped together, neurons are able to pass electrical signals through networks.
In 1854 George Boole released his theory known as Boolean algebra. This logic comprises of AND, OR, NOT commands and they could look like this: ( assuming the television is on and the radio is off )
Television is on = True
Radio is on = False
Radio is on AND Television is On = False
Radio is on OR Television is on = True
Pitts and McCulloch proposed a hypothesis of how neural networks make the brain work. The hypothesis showed that neurons could be seen as processors of binary numbers (1 meaning true, or 0 meaning false). This is also known as 'Parallel Computing'.
Using Boole's theory mathematician Walter Pitts and medical graduate Warren McCulloch wrote a paper on the neural network theory. It stated that the logical operations of True/False are created if the brains neurons does or does not send an impulse. This theory also proposed that memory could be defined as signals of a closed end loop of neurons. This theory of firing neurons is the basis on the artificial neural networks.
Pitts and McCulloch designed electronic replicas of neural networks, and showed how electronic networks could generate logical processes. They also stated that neural networks may, in the future, be able to learn, and recognize patterns. The results of their research and two books increased neural networks enthusiasm, and laboratories of computer simulated neurons were set up at a staggering rate.
Now the Top-Down Approach
The Top-Down approach deals with expert systems, expert systems are able to use logical factors to decide the best possibility. In the above example if the expert system was looking for a Non-Healthy, Main meal that tasted nice and isn't a vegetarian meal it's answer would be a Pizza.
This is human behaviour at 1000+ times the speed of a human brain. In the game of chess - a game requiring a high level of intelligence - a expert system can consider 126 million moves in 3 minutes, where a human can barely think of 5. The main difference between chess playing expert systems and humans is that expert system play their move using a highly logical outcome, coming from millions of possible moves. A human on the other hand is able to lean familiar and favourable board positions, and his/her move comes from the best barely 5.
These are just two of the theory's to creating AI, many of the other theory's are small variations of the other two, and some are completely unique. Other theories will become available with time, and more knowledge.