GenAI-GUIDE: a mixed-methods feasibility study of metacognitively guided generative AI in evidence-based nursing education

The rapid adoption of generative artificial intelligence (GenAI) in nursing education has outpaced guidance on how students should monitor, regulate, and justify its use in evidence-based nursing (EBN) learning. This study developed GenAI-GUIDE as a course-embedded teaching module to support students’ use of GenAI in EBN learning and evaluated its feasibility using implementation indicators and preliminary learning responses. This single-group, pretest–posttest mixed-methods feasibility study included 35 first-year students in a two-year baccalaureate nursing program in Taiwan, all of whom had prior nursing education experience. Informed by AI literacy, metacognition, and self-directed learning (SDL), the one-semester GenAI-GUIDE module combined AI-free and AI-supported stages with explicit AI-use boundaries, prompt refinement, risk awareness, and contribution disclosure. Students completed pretest and posttest measures of EBN beliefs and SDL, reported perceived human–AI contribution percentages across three EBN tasks, and submitted qualitative reflections. These data were used to examine feasibility and preliminary learning responses, rather than direct measures of EBN task performance or GenAI-use quality. Feasibility was evaluated using provisional benchmarks. Data were analyzed using Wilcoxon signed-rank and Friedman tests and inductive content analysis. All quantitative feasibility indicators met their provisional thresholds: recruitment was 71.4% (35/49), retention was 100.0% (35/35), and unit-level submission adherence was 95.2% (100/105). Findings describe within-group pre–post patterns rather than intervention effects. SDL scores increased significantly, whereas total EBN beliefs did not change significantly. Perceived AI contribution differed across PICO construction, literature searching, and evidence synthesis, but no pairwise comparison remained significant after Bonferroni correction. Six qualitative themes described students’ reflective GenAI use: teacher guidance, contribution disclosure, prompt structure, reappraisal of AI-generated content, tool-specific risk and verification needs, and collective monitoring through group discussion. GenAI-GUIDE was feasible to embed within an undergraduate EBN course. Future iterations should strengthen critical appraisal and clinical application, and controlled studies are needed to evaluate learning effects. This single-group pretest–posttest feasibility study was not registered in a public trials registry; therefore, no registration number is available.