Employee perceptions of AI adoption across service domains in a Finnish public health and social care organization: a cross-sectional mixed-methods study

Background: Empirical evidence on how employees across different service domains within a single multi-sector public organization perceive AI adoption is limited. This study, therefore, examined differences in self-reported AI usage, perceived competence, training needs, barriers, and motivations across four service domains within a Finnish public wellbeing services organization responsible for health and social services.

Methods: A cross-sectional, mixed-methods web-based survey was conducted among all employees of the organization in August–September 2025 (N = 437; response rate 10.3%; domain-level rates 6%–25%). The survey design and analytical lens drew on the Unified Theory of Acceptance and Use of Technology (UTAUT). Quantitative data were analyzed using descriptive statistics, chi-square tests, one-way ANOVA, and Spearman correlations. Open-ended responses were analyzed through reflexive thematic analysis.

Results : Self-reported AI usage varied across four service domains, from 78% among strategy and shared services respondents to 28% among elderly and disability services respondents (χ²(3) = 59.81, p < .001, Cramér’s V = 0.38). Perceived competence among AI users (M = 2.7/5) and motivation orientation did not differ across domains. Training need was high overall (M = 3.6/5), and the strongest bivariate association was between perceived competence and training need (ρ = −0.51). Among non-users, the highest-ranked barriers reflected facilitating conditions. Three of six qualitative themes converged on the same facilitating conditions UTAUT dimension.

Conclusions: Domain-level variation in generative AI adoption was consistent with an interpretation that structural facilitating conditions may shape whether training-focused interventions can translate into AI use in multi-sector public organizations.