Artificial Intelligence and Legal Reasoning: A Discussion of the Field and Gardner's Book

Dworkin 1977, 1985) offer insights valuable to AI, their jurisprudential analyses often raise more questions than they answer and their insights, couched in philosophical discourse, are difficult to harness computationally. Much of the knowledge used in legal reasoning is published, codified, and highly indexed. The legal system maintains extremely detailed records of its cases and commentary on them and except for the lowest-level courts, all cases are published and indexed commercially. For example, Shepard’s Citations records and updates all forward and backward pointers for cases; that is, for a given case, all the succeeding cases citing it as well as all the cases it cited. Some areas of the law are structured in ways that greatly facilitate knowledge acquisition and representation. For instance, some areas such as tax law have statutes so carefully drafted that they read almost like computer pseudocode or sets of logical propositions. In some areas, such as contract law, there are wellorganized efforts of legal scholars and practitioners to tease out, debate, and codify the principles and rules of the legal domain. For example, the Restatement of the Law, Second: Contracts (1981) is a compendium of principles, illustrative cases and hypotheticals, discussion, and case citations. Its 385 principles are abstractions of the holdings of thousands of contract cases. These were worked out in a series of meetings, held between 1962 and 1979 for the second Restatement. Certain sections of the first Restateegal reasoning is an intriguing field for the researcher in artificial intelligence because it demands that many deep and nettlesome problems, for instance, those of knowledge representation and analogy, be addressed head on. It presents a range of interesting reasoning skills, some of which seem tantalizingly tractable with currently well-understood methodologies such as those of expert systems, and others of which seem to subsume exceedingly hard problems such as natural language understanding. The law is an attractive domain for AI research for several reasons. First, it has a tradition of examining its own reasoning process. Second, its reasoning is stylized; in Anglo-American common law, one reasons according to stare decisis, or the doctrine of precedent, in which similar cases are to be decided similarly. Central aspects of such reasoning involve analogy and reasoning with cases. Third, much of the knowledge is readily accessible and some of it is very well structured and codified. Legal scholars and philosophers have a long tradition of grappling with the sort of issues—like the status of rules and exceptions—that are of interest to AI. Legal philosophy, or jurisprudence, seeks to examine the basis and workings of the legal system and to elucidate, among other things, the nature of legal concepts, rules and principles, the process of stare decisis, and the role of society and morality in law. While several legal scholars, (for example, Levi 1949; Llewellyn 1981; In this article, I discuss the emerging field of artificial intelligence and legal reasoning and review the new book by Anne v.d.L. Gardner, An Artificial Intelligence