Article | Published 31 March 2023
Applied Artificial Intelligence for Law: The Use of Expert Systems and Decision Trees in Legal Practice and the Potential of Artificial Neural Networks
Legal Issues Journal, Volume 9, Issue 2, January-June 2023
*Corresponding author. Email. firstname.lastname@example.org
Abstract. The literature on Artificial Intelligence (AI) as applied to legal practice can contain apparent paradoxes. On one hand, some researchers suggest that AI can transform law. On the other hand, others suggest that AI has more limited potential. This apparent tension can be understood by distinguishing different forms of AI technology. This paper reviews two contrasting types: First, it examines expert systems and their component ‘decision trees’; second, it considers artificial neural networks (ANNs). Finally, it reviews the use of the former within the legal sector and the potential of the latter. Expert systems have fixed ‘knowledge’ that is codified into computer systems by human experts. To some extent, expert systems enjoy wide legal- sector adoption–at least for the ‘decision tree’ part of the technology. Improvements for the legal sector can be made by enriching decision trees using methods from decision science and by using algorithmically generated decision trees. ANNs are quite different. They are vast networks of connections within software systems that ‘learn’ from rich data through exposure to it. The mechanism of ANNs is based on the principle by which neuronal connections in biological brains become stronger – via exposure and training. This paper finds that ANNs have significant potential but face more limited legal-sector adoption than decision trees. Other forms of AI exist but are beyond the scope of this paper.