作者
Heba Khalil,Nabeel AlYateem,Fatma Refaat Ahmed,Abedalmajeed Shajraw,Wegdan Bani‐Issa,Samia Kouki,Vidya Seshan,Jacqueline Maria Dias,Muhammad Subu,Khalil Yousef
摘要
Background: The rapid adoption of AI tools such as ChatGPT in nursing education raises concerns about the unverified application of AI-generated recommendations in clinical settings. Thus, this study aimed to examine the prevalence of unsafe ChatGPT use among nursing students, identify behavioral and demographic predictors, and assess the protective role of faculty guidance. Methods: A cross-sectional survey was conducted among nursing students in the UAE. Hierarchical binary logistic regression was used to examine predictors of self-reported unsafe ChatGPT use, including demographic variables, technology acceptance (perceived usefulness, frequency of use), and faculty guidance. Results: Approximately 39% of students reported applying unverified ChatGPT recommendations in clinical settings, with prevalence rising to 47.9% among 4th-year students. Seventy-four percent of students perceived ChatGPT as beneficial for academic performance. Regression analysis indicated that higher perceived usefulness$(\text{OR}=1.98,95 \% \text{CI}[1.42,2.76], p<.001)$and greater frequency of use$(\mathbf{O R}=\mathbf{1. 4 2, 9 5 \% C I}[\mathbf{1. 1 4, 1. 7 7}], \boldsymbol{p}$=. 001) increased the likelihood of unsafe use. Conversely, students who reported receiving guidance from faculty were protected from unsafe use$(\text{OR}=0.62,95 \% \text{CI}[0.46,0.83], p=. 001$). Conclusions: Unsafe ChatGPT use among nursing students is prevalent and primarily driven by behavioral factors, particularly perceived usefulness, and frequent use. Faculty guidance serves as a protective factor, underscoring the importance of mentorship and structured supervision. Nursing programs should integrate AI literacy, critical evaluation skills, and faculty support to ensure safe and effective AI use in education and clinical practice.