Bex, Floris; Prakken, Henry
Abstract: In this article, we discuss case decision predictors, algorithms that, given some features of a legal case predict the outcome of the case (i.e. the decision of the judge). Despite the enthusiasm in academia and the media about the opportunities offered by these so-called ‘robot judges’, there are also sceptical voices. In this article we will discuss possible uses of and objections against case prediction algorithms. In particular, we will look at whether, and if so how, such prediction algorithms can be used to support judges in their decision making process. We will look at which probabilities can be derived given an algorithm and its evaluation with a test set, and whether these probabilities can be applied by a rational judge
who has to decide a new case. We conclude that in practice such an application is almost never warranted. We further conclude that case decision predictors can only be useful in individual cases if they can give legal justifications for their predictions, and that only these legal justifications are what should matter for a judge.