Artificial intelligence in nursing practice – A Delphi study with ChatGPT

Abstract

BackgroundThe release of the open-source platform ChatGPT in November 2022 created massive interest in the possibilities of artificial intelligence–based chatbots. However, there is little knowledge about how artificial intelligence could be used in nursing practice. AimIn this study, we aimed to discover whether nurses' answers to nursing-related questions differ from those generated by artificial intelligence. MethodsA qualitative consensus group research design was used in this study. We conducted a Delphi study in three different rounds. For the first round, the instrument included 57 questions derived from 10 national nursing expert standards. In the second round, the nurses and ChatGPT were asked to rate the answers from the first round in order of relevance to everyday nursing. The third round included writing an assessment and a nursing handover based on two vignettes. ResultsThe study showed significant differences between the answers given by the group of nurses and those given by ChatGPT in all main categories. The results of the third round indicate that the answers from the nurses and ChatGPT did not differ, except for answers assigned to two categories. ConclusionsOur results indicate that ChatGPT generates answers that are comparable to those given by nurses. We assume that ChatGPT could be an enrichment at the beginning of the nursing process by summarizing and categorizing patient information. Furthermore, it could be helpful for nurses in the care planning phase. Given ChatGPT's ability to communicate in different languages, it might be a possible intervention for overcoming language barriers.