Algorithmic governance and due process in administrative procedures

Abstract

Introduction: This article examined algorithmic governance and due process in AI-assisted administrative decisions, highlighting risks of opacity, lack of explanation, and bias affecting social benefits, traffic, taxation, and public procurement. Objectives: To formulate and justify a minimum enforceable standard of transparency, explainability, and non-discrimination for the Ecuadorian context, aligned with international frameworks. Methodology: A systematic document review was conducted from 2016 to 2025 using a PRISMA protocol, independent double coding, and a 0–2 indicator matrix. The final corpus included 30 sources, balanced between Ecuador and international sources, and four critical sectors were analyzed. Results: Moderate procedural transparency was observed, generally higher outside Ecuador. Explainability, especially local explanations, constituted the most persistent gap. Non-discrimination remained largely declarative, with limited evidence and monitoring of bias in operation. Conclusions: Three complementary levers are proposed to close gaps and ensure auditable and fair decisions: standardized public records by system, with purpose, legal basis, data, metrics, responsible parties, and limits; institutionalization of systemic and local explanations with effective human oversight and avenues for challenge; and publishable impact assessments with periodic audits and redress mechanisms. The resulting standard strengthens motivation, accountability, and equal treatment.