Public perceptions of transparency in AI-driven decision-making in Sweden

Sammanfattning

The use of Artificial Intelligence (AI) in the Swedish public sector is growing, raising concerns about how ethical values like transparency, fairness, accountability, and privacy are upheld. While previous research stresses these values, little is known about how citizens perceive them in AI-driven decision-making. This thesis addresses that gap through two questions: (A) How do citizens perceive transparency, accountability, privacy, and fairness in AI-ADM in the public sector? (B) How do these perceptions vary by age, gender, occupation, and AI experience? A quantitative method was applied through an online survey distributed via snowball sampling. The survey included demographic questions and 18 Likert-scale items across the four value areas. Mean scores were calculated in Excel, and different statistical tests (e.g. Mann-Whitney, Kruskal-Wallis) were conducted in SPSS to examine demographic differences. One open-ended question provided qualitative insights that supported the interpretation of the quantitative results. Results show that accountability and transparency are seen as most important, while fairness and privacy received more mixed responses. Gender and occupation had a significant impact, while age and AI experience showed little influence. The findings suggest that public trust in AI relies on transparency and human oversight. Based on the results, this thesis provides several policy recommendations to ensure that the implementation of AI in the public sector maintains ethical values and public trust.