Why generative AI adoption stalls across public service domains: an AMO comparison of Finnish wellbeing and municipal technical services

Purpose This paper aims to examine how Finnish public employees’ perceived barriers to generative AI adoption differ between regional wellbeing services and municipal technical services and identify which ability-motivation-opportunity (AMO) theory dimensions and items pose the greatest constraints. Design/methodology/approach Two web-based surveys were conducted in September 2025. The wellbeing services survey targeted a single regional organization (n = 437); the technical services survey covered land-use planning and construction supervision across 104 municipalities (n = 218). Respondents reporting no prior AI use (n = 321) rated 19 barrier items on a 0–3 scale. Findings Opportunity barriers were the highest-rated AMO dimension in the wellbeing services sample and were significantly more intense there than in the technical services sample; the absence of organizational guidelines, weak peer support, and limited encouragement from colleagues or supervisors accounted for the largest sectoral differences. The technical services barrier profile was more evenly distributed across AMO dimensions. Comparable motivation composites masked an item-level reversal: doubts about generative AI accuracy and reliability were significantly stronger in technical services, where they ranked as the highest-rated barrier. Originality/value The paper offers comparative, employee-level evidence on perceived barriers to generative AI adoption among public-sector non-users. The findings suggest a contextual reading of AMO in which the weight of each dimension may depend on the configuration of public service production. Item-level disaggregation adds a second layer to this reading, since composite comparisons concealed the reliability reversal observed in technical services. Public organizations may benefit from diagnosing employee-level barrier profiles before designing generative AI implementation strategies.