Beyond the Internet: How Expert Systems Will Truly Transform Business

Charles R. Crawford, a former associate professor of computer science at
York University, is a computer-programming and mathematics consultant.

As this book says about the business world, “the future is always a
shifting array of possibilities driven by the self-fulfilling prophecies
of those who dream.” One prophecy that has driven many computer
software projects is that computers will someday gather and present
information in such a way that jobs like teaching and designing will
become much easier. Expert systems try to go one step further and
simplify decision-making in general. Smith’s thesis in Beyond the
Internet is that expert systems will be the next big software
application.

Chapter 1 focuses on technology as it describes the current business
environment, while Chapter 2 presents Smith’s view of expert systems
and how they differ from other computer software used in business. The
next eight chapters describe areas and associated problems that expert
systems can solve. Chapter 11 examines the consulting profession (which
Smith believes is the ideal way to market expert systems), while Chapter
12 describes the overall effects of expert systems on the business
environment.

Although he makes a good case for the need for expert systems, Smith is
not clear about how they will be designed and sold. His reasoning is
often muddled and contradictory. For example, he vividly describes the
problems caused by e-mail overload as well as the difficulties of using
existing filtering software, but he is less clear about how expert
systems can do a better job. In discussing product development, he
describes an “expert briefing program” that could “identify
underlying themes in human activity” and predict the success of a
particular product innovation. There is very little detail, however,
about how the program would identify these themes.

At one point, Smith refers to “spooky AI”—a reference that not
only seems to disparage artificial intelligence research but also
ignores that fact that serious AI research has found that the kind of
software he is proposing is very hard to design and test.