Catrina Denvir, Monash University, Tristan Fletcher, University College London, Jonathan Hoy, and Pascoe Pleasence, University College London, have published The Devil in the Detail: Mitigating the Constitutional & Rule of Law Risks Associated with the Use of Artificial Intelligence in the Legal Domain. Here is the abstract.
Over the last decade increased emphasis has been placed on the role that artificial intelligence (AI) will play in disrupting the practice of law. Although considerable attention has been given to the practical task of designing a computer to ‘think like a lawyer’, a number of related issues merit further inquiry. Of these, the risks that AI presents to the constitutionally protected procedural and substantive dimensions of justice deserve particular attention. In this paper, we consider the public and private application of AI in the administration of justice and the provision of legal services. We observe that the imposition of AI in certain legal contexts and settings has the potential to silence discourse between actors and agents, subvert the rule of law, and directly and indirectly threaten constitutional rights. In substantiating these observations, in Part I we begin by contextualizing recent developments in legal technology. Tracing the evolution of rule-based AI approaches through to modern data-driven techniques, in Part II we explore how AI systems have sought to represent law, drawing on the domains of: (a) judicial interpretation and reasoning; (b) bargaining and transacting, and; (c) enforcement and compliance, and we illustrate how these representations have been constrained by the AI approach used. In Part III we assess the use of AI in legal services, focusing specifically on implications that are posed in respect of the protection of constitutional rights and adherence to the rule of law. Finally, in Part IV we examine the pragmatic challenges that arise in balancing the risks and rewards of AI technologies in the legal domain, and we consider the issues that should shape and that are likely to shape use. We conclude by proposing the development of a ‘rule of legal AI’ designed to solidify the shared values that ought to govern future development in the field.
Download the article from SSRN at the link.
Comments