Inteligencia artificial y función administrativa en América Latina: riesgos jurídico-administrativos y lineamientos para prevenir la reproducción de desigualdades

Abstract

The article analyzes the implementation of artificial intelligence systems in the Latin American administrative function from a human rights approach, with the purpose of formulating legal- administrative guidelines aimed at preventing the reproduction of inequalities. The research is developed through a qualitative, documentary, comparative, and legal-hermeneutic methodology, focused on legal norms, public policies, international standards, and recent scholarship on artificial intelligence, public administration, and fundamental rights. The analysis argues that AI may support classification, prioritization, prediction, and recommendation tasks, but it cannot replace the legal formation of administrative will or the statement of reasons for administrative acts. In turn, the findings show that algorithmic opacity, bias, indirect discrimination, impairment of adversarial participation, hidden delegation, and evidentiary problems compromise substantive equality, due process, judicial review, and state liability. Finally, the article proposes guidelines based on prior delimitation of use, risk classification, data governance, transparency, auditing, impact assessment, reinforced reasoning, and human decision-making reservation.