Chapter 8 - Legal and Computer Rules: An Overview Inspired by Wittgenstein’s Remarks

Summary

The Concept of Rules between Law and AI

The concept of rules is a family resemblance concept. An initial survey of how such a concept is used across various disciplines reveals ‘a complicated network of similarities overlapping and criss-crossing’ (Wittgenstein 2009: PI § 66–67). A particularly relevant set of such family resemblance relations characterizes the field of Law and that of Artificial Intelligence, where the concept of rules plays a foundational role. This chapter will look at how the existence of family resemblance relations pertaining to the concept of rules has built a bridge between those respective fields of Law and AI and how this has laid the groundwork for the intersection between legal rules on the one hand and computer rules on the other.

It is significant that the initial research in the domain of AI and Law was set in motion by the consideration that ‘[t]here is more than a superficial resemblance between laws and computer programs’ (Niblett 1980, 8). Undoubtedly, the material conditions for the exploration of such a resemblance have been set by the increasing availability of ever more sophisticated computational machines. Yet, the emergence of the conditions of intelligibility for a crossing of the concepts of legal rules and computer rules should be understood as part of a larger, significant historical development of the general concept of rules. As demonstrated by historian Lorraine Daston, since the beginning of the modern age, the concept of ‘rules-as-algorithms’ has progressively gained ground (Daston 2022). The combination of such an algorithmic understanding of rules and the availability of more powerful computational machines made it possible to overcome the perception of the computer as a mere ‘mechanized storehouse’ (Kelso 1946) to stock and retrieve legal texts. Propelled by the developments in AI, the challenge became that of harnessing ‘the cognitive potential of computers for legal reasoning’ (Buchanan and Headrick 1970, 43). Thus began the efforts aimed at developing formal languages and programs to represent, in the form of computer rules, ‘not just the texts of the law, but the meaning of such text’ (Sergot 1991), that is, ‘the legal norms which they contain’ (Niblett 1980).