Human-AI interaction in high-stakes contexts: Child welfare professionals’ perceptions of AI integration

Abstract

Purpose: Understanding professionals’ early perceptions of artificialintelligence (AI) is particularly critical in high-stakes contexts, wheredecisions carry ethical and legal consequences. Such perceptions canhelp clarify acceptable-use boundaries, identify conditions for trust-worthy AI integration, and inform human-centered system design.Most prior research examines attitudes after organizational implemen-tation, leaving professionals’ perceptions in such settings underex-plored. This study addresses this gap by exploring child welfareprofessionals’ perceptions of AI in a context where individual-leveluse exists without formal institutional systems.Materials and Methods: Semi-structured interviews were conductedwith 23 child welfare professionals in Turkey. Data were analyzed usingreflexive thematic analysis in MAXQDA, within an abductive analyticapproach informed by the Unified Theory of Acceptance and Use ofTechnology.Results: Participants viewed AI as useful for administrative and pro-blem-solving tasks but expressed concerns about cognitive depen-dence and emotional unease. They drew firm boundaries around AI’srole in decision-making, emphasizing that ethical reasoning and con-text-sensitive judgment must remain human-led. Participants alsoidentified organizational implications of AI adoption, including theneed for institutional infrastructure and professional training.Discussion: These findings suggest that AI acceptance in high-stakeshuman services is shaped not only by perceived usefulness but also byethical concerns, professional values, and the perceived necessity ofpreserving human judgment. This underscores the importance ofcontext-sensitive approaches to AI integration in child welfare.Conclusion: Responsible AI integration in child welfare requires sus-tained collaboration among practitioners, organizations, and policy-makers. The study offers practical implications for frontlinepractitioners, organizational administrators, and national policydevelopment.