The complementary effects of algorithmic source and governmental context on citizens’ acceptance of algorithmic advice in government

The effective use of algorithmic decision-making in public service settings depends on citizens’ trust in both the algorithm and government that implements them. This study develops a model by integrating insights from trustworthy AI literature and social capital theory to explore the complex interplay between algorithmic factors and institutional contexts in shaping citizens’ adoption of algorithmic recommendations in government settings. We conducted a judge-advisor system experiment to examine citizens’ behavioural change influenced by algorithmic advice. The experiment manipulated two variables: the developer of the algorithm (private vs. public agency) and level of government utilising the algorithm (central vs. local government). The results indicated that citizens’ use of AI advice for behavioural change was highest when it was generated by a reputable algorithm and recommended by a high-trust government, demonstrating a complementary effect of the algorithmic source and the governmental context in shaping citizens’ use of AI recommendations.