@1. A Definition
When originally conceptualised in the 1970s, expert systems were thought of as an alternative to the human expert. Overwhelmed by the huge potential which the knowledge representation within computers gave them, scientista believed that as technology advanced it would become possible to replace the human expert with an electronic one. Developments in the field of Artificial Intelligence (AI) though, show that even more modest targets cannot be achieved and that the kind of expertise provided by an expert System can only be used as some kind of secondary support to the human one.
The definition of legai expert systems relies on the fulfilment of a number of criteria. According to Oskamp [Oskamp 1990,1992] a System has to contain the same knowledge of a particular legai domain as a human expert has. Thus, according to the classification of legai knowledge, there can be various categories of expert systems i.e. labour law, marmine law, etc. The pieces of knowledge have to be connected to each other so that the System can reason like a human expert. Simulation of the expertise humans possess can be achieved with the use of AI techniques. That results to better resources management. It also makes it possible for the System to reason on a particular conclusion by referrmg to legai sources it used [Reed 1990].
Early expert systems promised to cover large areas of various disciplines where the expertise could be represented in a computer executable program. Such programs could be copied and related applications could be widely distributed. The incentive was great and even non-computmg disciplines - such as law - felt the shock wave of this new idea and much enthusiasm for legai expert systems was generated [Leith 1991]. The idea that law was of a rule based nature just like expert systems created the false impression that ´it is just a matter of getting a group of lawyers off for a weekend togetherª (Leith [Leith 1991] quoting a senior figure of the uk government).Page 220
Many believed that a legai expert System could indeed become rcality. On the way though some realised that:
[expert systems]...are not as intellectually powerful as they were first thought; it began to be seen that thcy cannot really deal with ´common-senseª reasoning at ali. Further there seemed to be a real problem in moving the systems from laboratory over to the real application world when the umbilical cord between thc laboratory and the System in the real world ls cut, these systems oiten fall, and lie unused [Leith 1991].
@2. Drawbacks of the Application of Expert Systems
A typical example of a legai expert System is the ´Latent Damage Systemª which was launched in the uk market in 1988 [Capper et al. 1988]. The System provides expertise on a part of construction law. Although it is called an expert System it does not succeed in replacing a human expert completely. The System is more suitable for lawyers or users with a legai background than for use by non experts. The Latent Damage System rehes on the interpretation of the Latent Damage Act as given by its two authors. The developers of the Latent Damage System refer to it as an intelligent assistant for already knowledgeable users [Susskind et al. 1989].
Expert systems such as the Latent Damage System can cover a very small, not very demanding and quite confined field of law. They cannot interact with other related fields of law. Consequelltly, it is not possible to suggest that legai expert systems can be implemented in a way that is suitable to combine knowledge deriving from vanous sources, such as norms, procedure rules, empirical knowledge, etc. [Konstantinou et al. 1993]. Expert systems are not expected to be able to solve complicated cases and propose integrated solutions that involve the combination of knowledge from different legai fields.
Expert systems do not offer their users unprocessed legai material. In order to fulfil their purpose they contain legai information such as statutes and case law which have already been interpreted. Information provided by expert systems has been subjected to the human expert process which means that the impartiality of the raw legai information has been lost. Even when strictly controlied, the processed legai information cannot be quite accurate over a period of time or when conditions change. Consequently, legai expert systems need Constant attention in order to be kept up-to-date and distributed to their users.
Legai expert systems cannot be used in full scale real-life apphcations because they lack a reliable knowledge acquisition mechiamsm which wouldPage 221 permit them to be enriched with new data received from mteraction with users. Although lawyers' work is based to a great extent on knowledge that they have already acquired they are always in a position to obtain new knowledge if conditions become more demanding [Morison et al. 1992]. In die case of legai expert...