The prediction of legal outcomes and other legal domain related variables has served as the basis of a number of recent studies. While recent studies have estimated standardised variables and dichotomous outcomes such as the outcome of a judicial decision process, few studies have employed dichotomous data and categorical data to predict the basis of a legal defense strategy or the likelihood of trial success. Empirical research within the judicial sciences continues to employ a limited subset of empirical methods. This article reasserts the benefits of several artificial intelligence based non-parametric techniques that are better suited to the discipline than many of the common methods employed within the literature. The article considers the predictability of fair use defense within the U.S. during copyright infringement proceedings, and the likelihood of trial success.
TopIntroduction
The prediction of judicial processes has been the subject of much research in recent decades. There is a general acceptance of traditional economic methods within the sphere of legal research1 (Barker, 1996). The application of economic and econometric methods in the legal domain is ever increasing. Artificial intelligence (AI) methods are generating substantial interest with the legal community. Some might argue that the interest has been a long time coming given the advent of such technologies over three decades ago. Indeed the more practical advent of new methods, such as AI technologies in recent years has made such technologies more accessible to those within the legal domain2.
The advent of such technologies has not met with positivity from all practitioners. But the potential has resulted in a heightened sense of importance and a desire to develop greater familiarity that is palpable3. This occurrence is perhaps warranting some moderation of the evidenced euphoria about such methods in legal practice4. While some contend that AI methods have the potential to replace practitioners, the present article takes a contrarian view and argues that AI technologies shall serve to supplement traditional legal information sources, and inform trial strategy. There is also scope for the emergence of a secondary market for data-driven legal services.
Professor Art Cockfield moderated a recent discussion where participant innovator and legal graduand Addison Cameron Fuff offered some insightful comments. “I think the change is people being more proactive. Right now lawyers are very reactive; somebody has an issue, and a lawyer researches it using books and databases. There is an opportunity for software to make that first pass, to highlight new issues. When a new case comes out, you shouldn’t have to wait weeks for a newsletter. It should come into your inbox. That proactive aspect is something computers can deliver, because no lawyer on the planet could possibly read all of the cases, laws and regulations that come out. Depending on your scope, you could be talking municipal, provincial, federal, international…”5
Jordan Furlong in contributing to the same recent AI and Law dialogue asserts that; “we’re going to see the adoption of AI in the legal market, more broadly speaking, rather than in the legal profession for quite some time to come. Lawyers are sort of naturally disinclined, for cultural reasons, to disrupt the way they work and go about their jobs. Technology tends to generate that aversion”6. The present article offers an example of just how effective such methods and technologies might be in support of legal practitioners.
Within the legal domain the regulatory framework and its support structure are evolving, capturing more data and enhancing courts administration and judicial accountability.
Richardson (1989) posited the advantageousness of optimisation methodology, specifically of the economic flavour when discussing the role of analysis within the courts, the author's claims remain similarly valid to Artificial intelligence methods. Essentially courts and concerned with the allocation of resources and the behaviour of individuals. While somewhat reductionist there truth in the claim.
It is therefore logical to assume that the rules and models of sanction should be framed while having regard for the potential incentives and disincentives these rules and pronouncements create, and their likely impact on future resource allocations (Richardson, 1989). The current article considers the extant empirical research employed within the literature positing an alternative to the common logit methods employed within legal research.