|
ARTIFICIAL INTELLIGENCE
Artificial Intelligence or A.I. is a branch of computer sciences that does two main things:
- They study the computational requirements for reasoning, learning, perception and other such tasks.
- They develop systems to perform these tasks.
The field is very diverse, and with that it is also very difficult field to study. One can study the basic reasoning
of circuits (logic gates) and compare them to a human's brain, or one could look into a higher level of A.I. (e.g. ALICE)
that would include plans and goals. The factor which makes this field so challenging is the things which we take for
granted such as walking, opening and closing doors, are very difficult to program. For example if we want to walk
through walkway that contains a closed door, we open it, or use a key if it is locked. A robot may not even be able to
grab and turn the handle on the door, or unlock it if the door is locked, since robots have the inability to learn
complicated thing.
The ideas of A.I. originated thousands of years back in Greek mythology in which artefacts could reason automatically.
More recently in the 17th and 18th century clockmakers tried to make automations (see clockwork robots) that could
perform tasks on their own such as playing games or dancing. Generally A.I. is associated with circuits and computers
and in the 1950's after great strides in computer technology people were able to build systems to study and simulate
low-level artificial intelligence. Near the same time Alan Turning published a paper arguing the possibility of creating
an intelligent computing system, which led to the innovation of the Turing test (A test where a human and a computer
programmed to answer like a human which in separate rooms try to convince a second unknowing human that they are human.
The second human also asks the computer and first human questions). Early A.I. advanced quickly, there were programs
which could, solve equations, prove theorems in geometry and solve analytical. This advance unfortunately hit a wall in
the early 1960's; programmers of the time just could not use current programming methods to do more complicated tasks
and government funding was cut off. No one knew what was needed to make A.I. more knowledgeable. This led to research
in knowledge-based systems that were excellent for solving problems that required deep knowledge in a particular category.
Continued research showed that we needed to develop further methods for tasks, such as acquiring knowledge for systems to
use, dealing with incomplete or uncertain information and adapting to the situation. We must understand that A.I. as we
know it is a very young field. The term A.I. was only coined in 1956, and current methods of A.I. are far from perfect
but with time A.I. will advance to higher levels of intelligence.
|
