As stated previously, many Japanese companies that are not computer manufacturers have successfully deployed KBSs that are in day-to-day use and yielding a high ROI. Although most Japanese companies are reluctant to discuss ROI quantitatively, we gathered from some that they often require a two-year return of their investment in a system, and an investment of $500,000 or more would be considered large. On the other hand, they are quick to point out that there are other measures of success besides ROI: shorter turnaround, improved throughput, higher reliability, better individual decision making, etc.

Although several of the JCMs reported that the KBS business had flattened out, Mr. Yamanaka at Hitachi was optimistic about future growth for his company, and drew a rough sketch of estimated growth in four sectors: industrial, government, financial and retail (see Figure 7.1). The earliest major source of revenue for Hitachi has been in the financial area, but future growth will be relatively slow. Industrial applications are now starting to take off and will soon be the dominant source of revenues. Applications in the public sector will also grow at a healthy rate, but are starting from a low base. The retail sector is not expected to be investing in this technology in any significant way in the near future.

Figure 7.1. Forecast by Hitachi of Growth of Knowledge-Based Systems in Four Sectors
(Source: Hitachi, Private Communication)

Despite Hitachi's optimistic forecast of future growth in the deployment of KBSs, we found few signs of a multiplier effect in those companies where a successful application has been deployed. For example, JAL, with an excellent deployed application, is moving very cautiously towards implementing other applications. KBSs provide the highest benefit in very specific applications. Unfortunately, there is an inverse relation between specificity and size of market. Applications seem not to be transferable to other companies or easily even within companies. Contrast this with an airline reservation system, where thousands of clerks can use it as well as travel agents and other airlines. The lack of exponential growth may well mean that the cost of wildcatting may be too high. This cost is due to the limitations of first-generation tools which we heard about over and over again in Japan, as well as the lack of skills and AI knowledge.

Published: May 1993; WTEC Hyper-Librarian