The chatbot judge: reflections on the importance of Italian guarantees, taking a cue from a recent decision of the Colombian Constitutional Court

Abstract

The paper explores the legal and constitutional implications of using chatbots based on Large Language Models (LLMs), such as ChatGPT, in judicial proceedings. Focusing in particular on the Colombian Constitutional Court’s ruling T-323/2024, the authors reflect on the risks related to the use of AI in the decision-making phase of trials, including lack of transparency, judicial de-responsibilization, and violation of the right to defense. The study highlights the limitations of AI in ensuring rationality, transparency, and accountability in adjudication, emphasizing the need for a “human surplus” in judicial functions. It advocates for the use of AI as an administrative support tool, while excluding its legitimacy as an active subject in the production of judicial decisions.