AI Technologies in the Judiciary: Critical Appraisal of Large Language Models in Judicial Decision-making
Forthcoming in Regine Paul, Emma Carmel, and Jennifer Cobbe (eds), Handbook on Public Policy and AI (Edward Elgar Publishing)
28 Pages Posted: 10 Jan 2024
Date Written: December 18, 2023
Abstract
Using large language models (LLMs) in judicial decision-making carries social, organizational, and individual risks. This chapter provides an account of how the judiciary around the World, but especially in Latin America, have used LLMs to draft rulings or take decisions during court hearings and to problematize such uses by judges and their clerks and how they have justified these applications. The chapter argues that current LLMs cannot be regarded as trustworthy sources of information even if their capacity to produce language outputs mimics human ones in often seemingly convincing ways. Hence, LLMs should only be used – with the utmost care – when other more effective and safe judicial decision-making options are unavailable. Moreover, the chapter contends that the judiciary should promote digital literacy and informed, transparent, ethical, and responsible use of artificial intelligence tools to reap their potential benefits and prevent risks.
Keywords: large language models, ChatGPT, judicial, fundamental rights, Latin America
JEL Classification: K1, K4
Suggested Citation: Suggested Citation