Between caution and adaptation. Artificial intelligence in Polish social work practice

This article examines how Polish social work practitioners perceive artificial intelligence (AI), the extent of its adoption, and the hopes and concerns linked to its use. In December 2025, a nationwide Computer-Assisted Web Interview (CAWI) survey was conducted with 1,243 social workers and managers employed in social assistance centres. Findings show adoption is at an early stage: some respondents use generative tools, but for many AI remains marginal or absent from everyday practice. Perceived benefits focus on back-office support – saving time, organising and synthesising information, improving documentation quality – and thus freeing staff for relationship-based work with clients. Assessments of potential applications form a clear hierarchy: text support, translation, summarisation and information management are rated highest, whereas tools affecting direct client interactions (e.g. chatbots, automated diagnosis or needs assessment) raise the greatest concern. Key barriers are institutional and competence-related, including unclear rules, limited organisational readiness, and gaps in knowledge and training across diverse local welfare contexts.