Managing Information Technology

Chapter 6 • Managerial Support Systems 241

Explanation

Subsystem

User

Interface

Workspace Knowledge

Base

Knowledge

Acquisition

Subsystem

Inference

Engine

User

Knowledge

Engineer

FIGURE 6.6 Architecture of an Expert System

benefits as those of their more experienced counterparts
(Ko and Dennis, 2009). Collectively, these individual and
institutional factors facilitate the success of a KMS.

Artificial Intelligence

The idea of artificial intelligence (AI), the study of how to
make computers do things that are currently done better by
people, is well over 50 years old, but only in the last
30 years have computers become powerful enough to make
AI applications commercially attractive. AI research has
evolved into six separate but related areas; these are natural
languages, robotics, perceptive systems (vision and
hearing), genetic programming (also called evolutionary
design), expert systems, and neural networks.
The work in natural languages, primarily in
computer science departments in universities and in
vendor laboratories, is aimed at producing systems that
translate ordinary human instructions into a language that
computers can understand and execute. Robotics was
considered in the previous chapter. Perceptive systems
research involves creating machines possessing a visual
and/or aural perceptual ability that affects their physical
behavior. In other words, this research is aimed at creating
robots that can “see” or “hear” and react to what they see
or hear. With genetic programmingorevolutionary
design, the problem is divided into multiple segments, and
solutions to these segments are linked together in different
ways to breed new “child” solutions. After many genera-
tions of breeding, genetic programming might produce
results superior to anything devised by a human. Genetic
programming has been most useful in the design of
innovative products such as an energy-efficient halogen
light bulb that is much brighter than a standard halogen

bulb and a satellite support arm with a novel shape that

prevents vibrations from being transmitted along the truss.

The final two branches of AI are the ones most

relevant for managerial support. The expert systems

branch is concerned with building systems that incorporate

the decision-making logic of a human expert. A newer

branch of AI is neural networks, which is named after the

study of how the human nervous system works, but which

in fact uses statistical analysis to recognize patterns from

vast amounts of information by a process of adaptive

learning. Both of these branches of AI are described in

more detail in the following sections.

Expert Systems

How does one capture the logic of an expert in a computer

system? To design an expert system, a specialist known as

a knowledge engineer (a specially trained systems analyst)

works very closely with one or more experts in the area

under study. Knowledge engineers try to learn everything

they can about the way in which the expert makes deci-

sions. If one is trying to build an expert system for estate

planning, for example, the knowledge engineer works with

experienced estate planners to see how they do their job.

What the knowledge engineer has learned is then loaded

into the computer system, in a specialized format, in a

module called the knowledge base (see Figure 6.6). This

knowledge base contains both the inference rules that are

followed in decision making and the parameters, or facts,

relevant to the decision.

The other major pieces of an expert system are the

inference engine and the user interface. The inference

engine is a logical framework that automatically executes a

line of reasoning when supplied with the inference rules