Abstract
The laws that govern affluent clients and large institutions are numerous, intricate and applied by highly sophisticated practitioners. In this section of society, rules proliferate, lawsuits abound, and the cost of legal services grows much faster than the cost of living. For the bulk of the population, however, the situation is very different. Access to the courts may be open in principle. In practice, however, most people find their legal rights severely compromised by the cost of legal services, the baffling complications of existing rules and procedures, and the long, frustrating delays involved in bringing proceedings to a conclusion . . . There is far too much law for those who can afford it and far too little for those who cannot. No one can be satisfied with this state of affairs.
Derek Bok [5]
The American legal system1 is widely viewed as being in a state of crisis, plagued by excessive costs, long delays, and inconsistency leading to a growing lack of public confidence. One reason for this is the vast amount of information that must be collected and integrated in order for the legal system to function properly. In many traditional areas of law, evolving legal doctrines have led to uncertainty and increased litigation at a high cost to both individuals and society. And in discretionary areas such as sentencing, alimony awards, and welfare administration, evidence has shown a high degree of inconsistency in legal decision making, leading to public dissatisfaction and a growing demand for "determinate" rules.
In this article, we consider the potential of artificial intelligence to contribute to a more fair and efficient legal system. First, using the example of a middle income home buyer who was misled by the statements of a real estate broker, we show how a predictive expert system could help each side assess its legal position. If expert systems were reasonably accurate predictors, some disputes could be voluntarily settled that are now resolved by costly litigation, and many others could be settled more quickly. We then consider the process of discretionary decision making, using the example of a judge sentencing a criminal. We describe how diagnostic expert systems developed in the medical domain could be adapted to criminal sentencing, and describe a process by which this technology could be used—first to build a consensus on sentencing norms, and then to make those norms accessible.
In the ideal case, legal decisions are made after lengthy study and debate, recorded in published justifications, and later scrutinized in depth by other legal experts. In contrast to this ideal, most day-to-day legal decisions are made by municipal and state court judges, police officers, prosecuting attorneys, insurance claims adjusters, welfare administrators, social workers, and lawyers advising their clients on whether to settle or litigate. These decisions must often be made under severe pressures of limited time, money, and information. Expert systems can provide decision makers with tools to better understand, evaluate and disseminate their decisions. At the same time, it is important to reiterate that expert systems should not and cannot replace human judgement in the legal decision making process.
