Generative artificial intelligence in a blended problem-based surgical nursing course: a course-level needs assessment of student-reported risks and support needs

Generative artificial intelligence is increasingly used in student learning, but nursing education requires course-level guidance because students must connect digital information seeking with professional judgment, patient safety, and evidence-informed reasoning. This study described nursing students’ generative artificial intelligence use, perceived risks, and support needs in a blended problem-based Surgical Nursing course. We analyzed course-related modules from a broader course-team-developed online survey completed by all 305 eligible nursing students after a blended PBL Surgical Nursing reform course from late October to December 2025 (response rate, 100%). Valid responses were eligible course-cohort records retained after duplicate checking, identifier removal, and completion of the core modules. The questionnaire items were treated as exploratory survey indicators rather than as an externally validated standardized instrument. Descriptive frequencies, item diagnostics, data-quality checks, and supplementary exploratory association analyses were used. Generative artificial intelligence use was common: 261 students (85.6%) reported using it for learning at least three times weekly; this estimate was 246/290 (84.8%) after excluding frequency-time inconsistent responses and 246/305 (80.7%) under conservative recoding. Common tasks were homework assistance (74.4%), knowledge expansion (61.3%), course preparation (56.4%), experimental-design support (54.1%), literature searching and analysis (46.2%), clinical case analysis support (42.0%), and academic paper writing (35.4%). Students most often reported concerns about information accuracy (57.7%), variable generated-content quality (57.0%), terminology misunderstanding (55.4%), imprecise prompting (46.2%), and operational skills (27.2%). Support needs focused on advanced functions, discipline-specific applications, basic operation, case-based guidance, teacher-designed AI resources, critical-thinking guidance, digital teaching-tool use, and clinical-practice integration. The survey identifies guidance needs rather than evidence of generative artificial intelligence effectiveness. Findings support teacher-guided verification routines, nursing-specific artificial intelligence literacy training, prompt-formulation practice, basic operation guidance, and explicit course policies for responsible use.