El Derecho al Debido Proceso Puede Verse Amenazado Cuando Se Emplean Herramientas de Inteligencia Artificial Que Aún Son Desconocidas y Carecen de la Capacidad para Asumir Responsabilidades por Error o Impresiones en la Administración de Justicia

Abstract

This research article critically analyzes the risks that the implementation of artificial intelligence (AI) systems in the administration of justice poses to the right to due process, enshrined as a fundamental guarantee in Colombia's legal framework and international human rights instruments. Adopting a legal-constitutional approach, the study examines the theoretical framework of due process across three dimensions: (i) its anchoring in the Political Constitution of Colombia (Article 29), the "block of constitutionality" —which includes the American Convention on Human Rights and the International Covenant on Civil and Political Rights— as well as the jurisprudence of the Colombian Constitutional Court; (ii) the emerging regulatory landscape on AI, both in Colombia (legislative bills, resolutions from the Judicial Branch) and internationally (the European AI Regulation Proposal and OECD Guidelines); and (iii) the technical-legal challenges stemming from algorithmic opacity, the absence of civil or criminal liability for automated systems, and the inherent biases in training data. Using a qualitative methodology that combines legal, jurisprudential, and regulatory analysis, the article identifies specific tensions between the technical features of AI and the principles of due process. It references concrete cases of AI implementation in administrative judicial processes in Colombia, with a particular emphasis on the automation of decisions and their contrast with international human rights standards. The article concludes that while AI may optimize operational aspects of justice, its implementation without clear legal safeguards threatens essential procedural rights. It proposes a regulatory model based on four pillars: (a) explainable transparency as a requirement to validate automated judicial decisions; (b) establishing liability for errors or inaccuracies; (c) independent audits of AI systems involving human rights experts; and (d) the prohibition of decisions based solely on AI without human mediation. This work contributes to the academic debate on AI and fundamental rights by diagnosing potential constitutional risks and offering guidelines to harmonize technological innovation with the Rule of Law.