Student Engagement and Trust Calibration With Generative AI in a 5S Safety Assessment

Generative AI is increasingly used in engineering and technology education, but its role in safety education remains underexplored. This study examined undergraduate students’ ( N = 14) human-AI interaction during a 5S workplace safety assessment in a Human Factors course. Using a mixed-methods design, we analyzed User Engagement Scale–Short Form (UES-SF) responses and qualitative reflections after students independently completed a hazard identification task and compared findings with a parallel AI-generated assessment (ChatGPT-4o). Results revealed high perceived reward and value, minimal usability barriers, and mixed focused attention, suggesting task engagement without deep immersion. Qualitatively, students described AI as expanding hazard awareness, improving analytical structure, and accelerating information retrieval, while consistently emphasizing the necessity of human oversight and cautioning against overreliance. Integrated findings position generative AI as a human-in-the-loop decision-support tool: effective for scaffolding systematic reasoning and broadening hazard recognition, but insufficient as a substitute for domain-informed judgment.