Abstract
<title>Abstract</title> <p>Background Large language models (LLMs) such as DeepSeek, Doubao, and ChatGPT are increasingly being used by undergraduate nursing students for learning, clinical preparation, and academic writing. However, little is known about how students actually use these tools, the specific benefits and challenges they encounter, their lived experiences during use, or their perceptions of academic integrity in the context of LLM use. Aim To explore the current status, experiences, and needs of undergraduate nursing students using large language models, and to provide empirical evidence for integrating AI technologies into nursing education. Design A qualitative descriptive study. Participants Fifteen undergraduate nursing students with regular LLM use experience (at least twice per week over the preceding six months). Methodology Data were collected through semi-structured, face-to-face interviews using an interview guide that was pilot-tested with a small group of nursing students and refined prior to the formal interviews. All interview data were transcribed verbatim and analysed using Colaizzi's seven-step method. Results Four main themes emerged: (1) Current patterns of LLM use—students employed diverse tools across multiple scenarios using rational strategies. (2) Benefits and dilemmas of the user experience—benefits included improved efficiency, convenience, and interactive engagement; dilemmas encompassed insufficient accuracy, content redundancy, knowledge lag, poor alignment with textbooks, inability to recognise nuanced needs, and cost barriers. (3) Perceptions of academic ethics—students distinguished legitimate assistance from cheating, while expressing concerns about over-dependence and erosion of critical thinking. (4) Needs for tool optimisation and AI literacy—students called for nursing-specific knowledge bases, source referencing, multimodal functions (images, videos), free or affordable access, and the addition of systematic training in the use of LLMs. Conclusion Undergraduate nursing students use LLMs pragmatically and strategically, achieving clear gains in learning efficiency despite notable limitations in accuracy, professional relevance, and accessibility. Students show some ethical awareness but worry about over-dependence on these tools. Accordingly, nursing education should establish discipline-specific usage guidelines, develop nursing-tailored LLMs, and offer AI literacy courses. Such measures would foster deeper integration of intelligent technologies into nursing education and cultivate nursing professionals with both clinical competence and digital literacy. Clinical trial registration Not applicable.</p>