A Software Architecture Model for Reducing Citizen–Government Trust Deficit in AI-Driven e-Government

Abstract

The integration of AI into many platforms allows for automation and robotic Process Automation (RPA), predictive analytics, and data-driven decision-making, each of which streamlines operations. Despite the progress, citizen trust in AI-empowered government systems is fragile due to issues of algorithmic opacity, data privacy vulnerabilities, bias and weak accountability mechanisms. While existing research addresses trustworthy AI broadly through ethical principles, governance frameworks, and regulatory approaches, little has been written on operationalizing trust in software architecture. This gap is currently addressed by the this paper, which proposes a software architectural model intended to mitigate the citizen–government trust deficit in AI-enabled egovernment systems. Through a Design Science Research approach, trust-related requirements are derived from literature and transformed into architectural entities integrated within distinct layers of the system. This paper proposes a novel framework that combines transparencyby-design, explainable AI, accountability services and data governance with citizen engagement approaches. The paper makes an important contribution by showing that trust can be baked into the system as a fundamental architectural feature rather than considered entirely at the policy or governance level.