Interviews with 20 Korean nursing applicants and educators find AI competency assessments efficient but opaque in criteria and weak on nursing's human core

The study identified four themes: doubts about the evaluation method and criteria, efficiency of AI-based competency assessments, challenges in preparing for AI-based competency assessments, and improvements and alternative approaches to nurse selection, synthesized from eight subthemes and 32 codes. Prior research on AI recruitment tools focused mainly on hiring managers, and although AICAs have been used in Korean nurse selection for about 4 years, their effectiveness in the nursing context had not been examined; this study is the first to include both nursing applicants and nurse educators (professors and nurse managers). Based on five semistructured focus group interviews in South Korea between February and August 2024 (two with applicants, three with educators, three to five participants each), totaling 20 participants, using conventional content analysis and the COREQ checklist, with data saturation reached at the fifth group interview and member checking used to confirm themes.

Participants recognized AICA's efficiency value: freedom from time and location constraints, easier handling of multiple job searches alongside final-semester classes, multifaceted screening of concentration, agility, and personality, and reduced personnel costs and fatigue in interview evaluation. Extends the efficiency advantages of AI recruitment from the manager perspective to the shared experiences of nursing applicants and educators, and specifies dimensions AICA can assess that short face-to-face interviews struggle to capture. Interview statements from applicants, professors, and managers, such as an applicant noting participation at any time and place to avoid skipping class, and a manager noting reduced human resource evaluation burden allowed more focus on nursing duties.

Participants questioned AICA's alignment with core nursing values: automated interviews were seen as unable to reflect person-centered nursing, therapeutic communication, and empathy, and opaque evaluation criteria left both those who passed and those who failed unclear about the reasons. Concretely presents the tension between AI selection tools and the humanistic nature of nursing in the nursing context, complementing prior AI recruitment research centered on manager perspectives. Direct quotations from applicants, professors, and managers, such as an applicant calling it a contradiction to use AI to select nurses who care for patients, a professor describing AI evaluation as mechanical and unable to reveal students' warm human side, and a manager noting the hospital's mission was not reflected in evaluation criteria.

Participants pointed to fairness risks in AICA preparation from external environmental factors and digital readiness gaps, and proposed improvements: developing nursing-specific AI programs, building algorithms on big data, periodically reviewing evaluation methods, using AICA results to support face-to-face interviews, providing clear guidelines, and strengthening nursing students' digital literacy education. Expands the fairness issue beyond the algorithm itself to external conditions such as network stability, lighting, noise, devices, and information-support gaps, and offers concrete improvement paths proposed jointly by applicants, professors, and managers. Applicants described the system crashing for three hours due to simultaneous connections and logging in early in the morning and seeking a well-lit, soundproof internet cafe; professors and managers worried about gaps between students who can freely use information technology and those who cannot.

The study addresses the recruitment context of tertiary hospitals in South Korea that have implemented AICAs, and is suited to understanding how nursing applicants and nurse educators (professors and nurse managers) perceive AI-based competency assessments, offering improvement directions for medical institutions and nursing schools: developing nursing-specific competency models and algorithms, establishing transparent and consistent evaluation criteria, using AICA results to support face-to-face interviews, and providing standardized preparation guidelines and mock assessments to strengthen digital literacy. The authors recommend collaboration between medical institutions and nursing and AI experts, and that policymakers and hospital administrators establish standardized guidelines and robust technical infrastructure.

The authors note the study is limited to specific Korean cultural and institutional settings, all participants had recent AICA experience so the diversity of perspectives was limited, the qualitative interview approach cannot objectively assess AICA effectiveness, and findings may be time-sensitive given the rapid evolution of AI technologies. In addition, AICA's impact on nurse adaptation and turnover has not yet been analyzed, the evaluation criteria themselves remain unclear, and how nursing-specific algorithms and standardized guidelines should be developed and validated remains an open question.