Intelligent systems: an emerging business tool. (Technology).

Artificial intelligence, long the stuff of science fiction films and novels, is beginning to play a significant role in the day-to-day lives of Canadians. "Intelligent systems" are being used by builders to make more accurate estimates, by lending institutions to predict the potential for business bankruptcy, and by marketers to assist in classifying and segmenting their customer base. In fact, intelligent systems are now in use in most sectors of the Canadian economy.

Intelligent systems have been defined as systems able to "emulate the human ability to perceive, reason, make decisions, and act." They enable machines and devices to anticipate requirements and deal with environments that are complex, unknown, and unpredictable.

Intelligent systems were first proposed in the late 1950s, but their practical applications have mainly developed since the 1980s. They were originally used in traditional industries such as manufacturing, mining, and forestry to enable the automation of routine or dangerous tasks and to improve process quality.

One area in which intelligent systems are finding application is in assessing the "health" of small and medium enterprises (SMEs). Since SMEs now provide over 90 per cent of the jobs in Canada, considerable efforts have been made to determine the key factors critical to their success. By transferring this knowledge to the entrepreneur, it is anticipated that the SME sector will be strengthened, resulting in increased employment and wealth for the country.

One method of creating this information involves assessing a company's weaknesses and strengths by comparing them to the best practices of a large number of successful companies. A business analyst usually does this assessment, but the process can be very time consuming and is sometimes subjective. Now, a type of intelligent system called a neural network can be "taught" how to undertake this kind of business analysis.

A neural network is basically a mathematical model of brain functions built on a computer. With sufficient complexity, a neural system can be enabled to learn, so that all the processes involved in assessing a problem do not have to be known in advance.

This application of neural technology is developed by feeding hundreds of business analyses and recommendations done by human experts into a computer. The neural system does not try to explicitly figure out the criterion that human beings used to make these evaluations. Instead, it "looks" for patterns: when certain inputs repeatedly yield certain recommendations, the neural network learns the patterns and can make similar recommendations from new inputs.

One advantage of the neural technology is its speed: A business could go online and fill out a detailed questionnaire which would be automatically assessed by an intelligent system, resulting in an immediate set of prioritized recommendations for improvements to business practices. Another advantage is increased objectivity: While a human analyst might be influenced by various subjective factors, anything from gender to style of dress, a neural network is oblivious to such feelings. Intelligent systems do not eliminate human involvement, however, since the technology is often used to help expert analysts check their assessment of a business against a more objective standard.

The applications of intelligent systems to assist businesses to automate complex and time consuming tasks is spreading to a variety of industries, from agriculture, manufacturing, transportation and communications, to financial analysis, electric power generation, and health. Some practical applications include: methods of sorting seeds to identify the best seeds and eliminate damaged ones; predicting the number of people likely to take a given flight, to avoid over- or under-booking airline reservations; and developing adaptive control systems to optimize the conversion of a wide variety of different vehicles to natural gas.

In Canada, Precarn Incorporated, a national consortium of corporations, research institutes and government partners working within the intelligent systems industry, promotes the application of intelligent systems such as neural networks. More information about Precarn and intelligent systems can be found at www.precarn.ca.

Kim Young is a Principal Research Engineer with the Saskatchewan Research Council