The current state of AI, one man's opinion
Roger C. Schank · American Association for Artificial Intelligence eBooks · 1989
M IS IN A SERIOUS STATE OF DISRUPTION The field has always been somewhat fragmented as a result of the history of its development. Different laborat.ories have tended to keep their own best students, and have inbred to the point where the theories and programs invented at one laboratory have often had very little relationship to those developed at another If t,his parallel development had involved only theories and programs, the problem might not have been so acute. However: it has usually been the case that even the idea of what tropics are significant to work on has developed in parallel. AI would have a fairly difficult time justifying itself as a scientific discipline with a coherent developmental pattern Wit,hout a coherent methodology and clearly defined goals, sometimes it seems that all AI people really agree on is that computer models that replicate intelligent behavior are worth working on (in some cases). Recently, this haphazard situation has goMen worse. It is natural that many AI people have a need to create programs that are used by somebody for some real purpose This need has begun to express itself in a way that may alter the nature of the field profoundly. The advent of the business world’s interest in AI, together with the frustration that most AI people feel about legitimizing what they do in the eyes of the world have changed the face of AI, but, unfortunately, this change has not necessarily been for the better. In this article I wish to address some of the problems that confront AI. I am giving, no doubt, what amounts to no more than one man’s opinion. It is my hope, in expressing these opinions, that the issues begin to be discussed in some public forum. I will attempt to start this debate by answering some questions about the field that have been posed t.o me over time. In some cases, what follows are questions t,hat I have simply posed to myself.