Transparencia de la inteligencia artificial en la administración pública: una revisión de estándares internacionales

This article compares soft law standards applicable to systems automated by artificial intelligence (AI) in public administration, with the purpose of promoting governance with reliable, transparent, accountable and people-centered algorithms. The methodology is based on a formal dogmatic comparative review to propose principles applicable to this technology with a descriptive and propositional approach. The study concludes that, despite the rapid evolution, the discussion predominantly focuses on technical aspects of data protection, neglecting keys such as protecting the right to access, algorithmic explanation, and risks to guarantor democracies. The study contextualizes five essential principles of algorithmic transparency of public administration. It emphasizes that the limitations of these studies stem from uncertainty regarding legislative adoption, primarily because they lack binding effects, change rapidly, and are not a direct source of domestic law. The main contribution of the article is to provide a global vision and an attempt to harmonize them, urging political actors to carefully examine the comparative experience in order to promote the necessary regulations to increase citizens’ trust in systems automated by algorithms, and stricter limits to artificial intelligence in public administration.