Rebuilding Trust in Fragile Democracies: Algorithmic Accountability and Citizen Perceptions of AI in Sri Lankan Digital Public Services

Abstract

This study investigates how algorithmic accountability and citizen perceptions shape trust in AI-driven public services within evolving post-crisis governance context in Sri Lanka. Based on 45 semi-structured interviews with government officials, technocrats, and citizens, the research identifies five interrelated challenges: institutional distrust, algorithmic opacity, digital exclusion, cultural disconnect, and the demand for human-mediated accountability. These themes reveal that citizen trust in digital governance is not solely a function of system transparency or efficiency but is profoundly influenced by historical marginalization, socio-cultural conditions, and procedural fairness. Hence, this study advances the concept of Trust-Centric AI Design, a framework grounded in institutional trust theory, which integrates explainability, grievance redressal, and participatory co-design. The findings contribute to regional debates on inclusive e-governance and emphasize the limitations of Global North-derived models in fragile democracies. As Sri Lanka expands its digital transformation agenda, this research offers a context-sensitive roadmap for ethically resilient algorithmic governance.