Justice and Generative AI: The Constitutional Challenges

ABSTRACT This article investigates some of the multiple intersections between artificial intelligence (AI) and law by focusing on the implications of using the latest forms of generative AI in judicial processes. After introducing the subcategory of Large Language Models, the author examines the amateur use of ChatGPT in some recent cases, highlighting the potential benefits, such as improved efficiency and effectiveness of decisions, and the risks concerning transparency, personal-data protection, judicial independence, and the right to a fair trial. The article then explores some institutional documents that govern the development of digital justice in different jurisdictions (Court of Justice of the European Union, European Court of Human Rights, England and Wales Courts) and addresses AIassisted human decision-making. The last part discusses whether we are at the dawn of a new form of justice, emphasising balancing technological advances with protecting fundamental rights. “The law is not a machine and the judges not machine-tenders” (J. Frank, Law and the modern mind, London, Transaction Publishers, 1930).

KEYWORDS Digital justice – AI Act – Generative AI – Predictive Justice – Constitutionalism

TABLE OF CONTENTS 1. AI and justice. – 2. Applications of AI in the legal field. – 2.1. LLMs. – 2.2. The uses of LLMs in the field of law, with a first impression on them. – 3. Technical problems arising from the use of LLMs in the legal field. – 4. From the Ethical Charter of the CEPEJ to Documents and Guidelines on the Use of GenAI in Judicial Activities. – 5. The AI Act. – 6. Constitutional profiles relating to the use of LLM in the judicial process