Petroleum Applications: How Do We Realise the Potentia ?

This paper is an edited transcript of the opening keynote lecture at the European Conference on Artificial Intelligence in Petroleum, Exploration and Production, October 1991. This talk addresses the question : given so much effort has gone into developing expert systems and developing the technology, why are there so few applications? What are the reasons, and more importantly, what can we do about it? In this paper, I will give several examples of working expert systems my company has developed, and what we think we have done right in each situation. I will talk about one or two problems or failures that have occurred along the way. I'll mix examples of real systems with lessons and recommendations, things you must be very careful of if you want to build successful applications. This will include one of the largest real time expert systems in Britain, and probably one of the most widely used expert system applications in the Whole engineering field, which is one that my company has developed with over 450 users just now.