CHAPTER I: GENERAL CONSIDERATIONS ON ARTIFICIAL INTELLIGENCE
AND EXPERT SYSTEM.
In this chapter, I'm going to deal with definition of
artificial intelligence, historical overview of artificial intelligence, fields
of the artificial intelligence, intelligent agents, advantages of an expert
system, characteristics of an expert system, and elements of an expert
system.
I.1. Artificial intelligence
I.1.1. definition
Several authors have defined the artificial intelligence
according to their understanding because there is not any consensus of the term
`'artificial intelligence». Here are some definitions that we had found in
the literature:
ü The study of mental faculties using models of the
calculative type. CHARNIAK et al (1985);
ü Conception of intelligent agents. POOLE et al (1998)
ü The automation of the activities associated with the
human reasoning, such as making the decision, solving problems, and the
training. BELLMAN (1978);
ü Construction of the machines which can be like human
being (vision, comprehension...)
ü The artificial intelligence is a whole of techniques
aiming to approach the human reasoning.
As we can notice it, these definitions agree each other on the
fact that the objective of the AI is to create intelligent systems, but they
are different in their way of defining the intelligence. Some are focused on
the behavior of the system, while others consider that what is important is the
working procedure (reasoning) of the system.
A second distinction can be made between those who define the
intelligence starting from the human being and those who do not refer to humans
but to a general standard of rationality. Thus, we can indicate four ways to
see the artificial intelligence:
- To create systems which behave like the human beings. This
operational definition of the AI was promoted by ALAN TURING, who introduced
his famous "test of Turing" according to which a machine is regarded as
intelligent if it can converse in the way that the questions (humans) cannot
distinguish it from a human being.
- To create systems which think like human beings. If we agree
with this second definition, that implies that the AI is an applied science,
because it is necessary to include/understand, as a preliminary the way in
which the human one thinks (if not, how to know if a machine thinks like a
man?) and then to assess the systems compared to their similarities with the
human reasoning.
- To create systems which think rationally. According to this
definition, the system must reason in a rational way, i.e. following the
logical rules. This approach can be criticized because it seems that some
capacities (perception, for example) are not easily expressible in logic. This
standard of rationality cannot be reached in practice because current
technology does not allow to realize complex calculations.
- To create systems which possess rational behaviors. This
last definition of the AI relates to the development of the agents which act
for better so as to satisfy their objectives. It is worth noticing that this
definition is more general than the preceding one.
| 
