Expert Systems: How Far Can They Go? Part One

done. In the middle there are a variety of agnostics and partial believers who aren’t sure, or are interested and want to discuss the questions but don’t have a firm view. Then, on the other side you have the TV evangelists who say, “Send in your money and you shall receive miracles.” I think there’s another similarity sometimes, which is that the discussions tend to get couched in terms of what it’s going to be like in the sweet bye and bye. That is: “Can you imagine that at any day in the future it might be possible that a computer could do this or couldn’t do that?” It is easy to lose touch with the reality of what people are actually doing—with what kind of sense it makes now and what kind of sense it is likely to make in the foreseeable future. So, my response to those comments was “Yes, indeed, I don’t want to make this a religious discussion. I think that it will be more useful to everybody to focus on a more specific set of questions that are more practical.” In this panel, we will take on questions about what expert systems really are doing and can do. By that I don’t mean focusing on just the current systems. I don’t think it’s particularly useful to pick apart the weaknesses of a particular knowledgebased system or a particular rule formalism. We need to take a broader view, covering the body of techniques, of which particular systems are examples, and anticipating to some degree where they’re going to go. Looking at where things stand now, it is interesting to see how they are different from what they were two to three years ago, which was again different from a few years before that. There has been a huge wave of enthuWe are in the midst of a great wave of enthusiasm about the potential for expert systems in every area of human life and work. There is no agreement, however, as to just how much they can do, and where they will run into fundamental limits. The intent of this panel is to present and discuss some basic questions as to what expert systems can really be expected to do: • What is the nature of the problem domains in which expert systems are likely to succeed and those in which they will not? Are there domains in which their use might be dangerous? • How will their performance compare with that of human experts in the domain? Are there different facets of expertise that are not amenable to programming? How can human and machine expertise best be combined? • To what extent can we count on rule-based systems for “flexibility” in dealing with unexpected situations? How reliable will such systems be in cases where the programmers (or knowledge engineers) did not anticipate significant possibilities? • How can a knowledge base be subjected to standards of accountability? Who is responsible for what an expert system contains and what it does?