Using Issues to Explain Legal Decisions

Abstract

The need to explain the output from Machine Learning systems designed to predict the outcomes of legal cases has led to a renewed interest in the explanations offered by traditional AI and Law systems, especially those using factor based reasoning and precedent cases. In this paper we consider what sort of explanations we should expect from such systems, with a particular focus on the structure that can be provided by the use of issues in cases.

Argument Schemes for Factor Ascription

“…Neither exigency nor privacy can provide a reason to decide the case by themselves: they must be considered together. Thus the factor is something like SufficientRespectforPriva-cyGivenExigency, and we can picture the dimensions as the x-and y-axes and precedents determining a line separating the area where the factor applies from where it does not, as discussed in [7] and [6]. Again, this line will be constrained by previous decisions as to the applicability of the factor.…”

“…In Steven's prime example in [23] the (hypothetical) case turns on whether a kindergarten teacher should be considered sufficiently analogous to the mother in the case of Dillon v Legg 5 to receive compensation for witnessing an injury to a child. In Dillon it was held that one factor was whether the victim and the emotional sufferer had "a close relationship" 6 . The case turns on the point because the other pro-claimant factors are in place, and so all that is at issue is whether being a kindergarten teacher enables the factor CloseRelationship, introduced to describe the mother-son relation, to be applied.…”

“…The next scheme concerns trade-offs between two dimensions, as described in [6]. For example in the US Fourth Amendment domain there is a trade-off between being able to enforce the law and respect for privacy [9].…”

“…The factor involves balancing these two concerns and is something like "Sufficient respect for privacy while enabling enforcement". The idea in [6] is that a line, e.g a.D 1 + b.D 2 + c = 0, can be fitted to the precedents 8 , separating the pro-plaintiff and pro-defendant regions of the case space. In the equation a and b are the coefficients of the variables D 1 and D 2 , representing the values on the two dimensions.…”

“…Dillon v Legg 68 Cal. 2d 728 (1968) 6. The Court had been deliberately vague, saying that it "cannot now predetermine the defendant's obligation in every situation by a fixed category; no immutable rule can establish the extent of that obligation for every circumstance in the future."…”