In this issue of AI and the Net, we briefly discuss a press release on a report from Business Communications Company, which if correct, estimates the AI market for 2002 at $11.9 billion. Having followed the AI industry for over 20 years and not having reviewed the actual report, I was curious how they arrived at those estimates. On the surface, vendors that supply AI technology are not currently in a market with a size anywhere near that, even with some products having a price tag in the $100,000's. There currently isn't the volume. The reopening of DARPA AI and cognitive research (see PC AI 16.5) will increase the AI market somewhat. However, if we look at the large vertical applications based on AI technology, such as business rules, Customer Relationship Management (CRM) and bank fraud, just to list a few, these numbers are very reasonable. Adding academic and science research and military applications, these numbers indeed do appear to be very reasonable. It also illustrates something else that seems to escape much of the general press and that is the value and acceptance of AI technology in the business mainstream. Major business application suppliers, such as IBM, Platinum, Blaze, just to name a few, have fully embraced the expert system technology and have customized it for their customers. Just like object oriented development, expert systems are another successful AI technology that fails to be recognized for what it is - fundamental AI.

       This issue has another PC AI exclusive written by Don Barker, a long-time writer and columnist for PC AI. Don spent the past two and a half years researching the Microsoft Research Division (MRD) preparing to write a number of articles and a possible book on the history and accomplishments of MRD. As has been mentioned a number of times in these pages, Microsoft has been hiring the best AI researchers and developers in numerous AI related fields. As very few writers have access to this type of information, we are excited that Don has written this second article, based on his research. This article, on Microsoft's new .Net Speech Platform, first presents a brief overview of the origins and the history of speech recognition at Microsoft. It then covers Kokanee, the code name for this project.

       This issue has some very interesting and informative articles. Girish Keshav Palshikar looks at how rule classification can improve the quality of Rule-Bases. In his article, he uses Prolog's ability to do logic programming to simplify the task, although the concepts are valid independent of language. For those that are not familiar with Prolog, or that want a quick refresher, Thomas Linder Puls offers a quick overview of Prolog fundamentals. Bob Nisbet proposes a link between the philosophical studies of the nature of being and CRM studies of customer response. Although this may sound very abstract, it really ties Abstract Modeling of customer behavior to well-focused data mining - all very practical and fundamental in business practices. The Robotics column revisits some old friends, such as Kismet and ASIMO, and introduces some new ones. The AI-Q challenge this issue is a relatively tough one focusing on natural language processing