[an error occurred while processing this directive]
This paper has pilfered from a diverse literature in order to overview, in a single paper, the rapidly evolving area of software agents. Only one other paper (Wooldridge & Jennings, 1995a) has attempted a similar extensive review of this area, which they do from a theories, architecture and languages angle. In this paper, we have overviewed the same area from the viewpoint of the clear diversity of agents being investigated in universities and research laboratories worldwide. We hope it provides a useful contribution to understanding this exciting field of software agents. However, we conclude this paper with the following, arguably controversial and/or casual, postscript.
The word ëagentí is currently in vogue in the popular computing press as it is within the artificial intelligence (AI) and computer science communities. It has become a buzzword because it is both a technical concept and a metaphor. However, its rampant use should conjure up the problems faced with other flamboyant titles including ëartificial intelligenceí itself: far too ambitious claims precede the real technical work that follows; the subject tends to attract dilettantes who just ëjump on the bandwagoní in order to make a "quick buck" or for some other ulterior motives; and coupled with another contentious word such as ëintelligentí as in ëintelligent agentí, prima facie at least, it generates unrealistic expectations of the state-of-the-art. These exaggerated claims, over-expectations and the ëoversellí eventually drown out the real (and sometimes excellent) achievements, which never match the hype. In short, much is promised but little is delivered! The aftermath of all this is usually quite predictable: interest in the area soon wanes and the research funding, gradually or rapidly, disappears. Eventually, the dilettantes leave in search for ëgreener pasturesí and leave the real researchers to pick up the pieces. Is this the fate for agents research - a passing fad? Hopefully not!
There are several things which the serious agent researchers can do. Firstly, they can drop the ëintelligentí in intelligent agents as we have done in the title of this paper: its connotation, and hence expectations, are much less. Secondly, they could attempt to ensure, where possible, that dilettantes do not publish articles on agents in the popular press, at least not until it has been reviewed by someone whose interest in the area is more than superficial. This is sometimes possible because some experts usually get asked to review such articles before they go to press. Thirdly, they (i.e. experts) should not themselves engage in overselling the domain and, lastly, they must always be critical of the progress in the area in order to provide a more realistic appraisal of the state-of-the-art. In this overview paper, we have attempted to abide by the principles of this doctrine. It is up to the reader to judge how successful we have been in meeting these principles in this paper. [an error occurred while processing this directive]