An attempt to analyze interview articles with registered dietitians in Japan by humans and generative AI

Abstract

　This study aimed to evaluate the text analysis capabilities of generative artificial intelligence (GenAI) by analyzing interviews with registered dietitians on the websites of hospitals and healthcare facilities. These interview responses were simultaneously read by humans and GenAI, and the results were compared. This approach enabled us to gain valuable insights into the current status of registered dietitians. Findings revealed that GenAI was able to understand the context of the interview responses to a certain extent and generate text that effectively expressed the participants' opinions. However, when reading the text, GenAI tended to make excessive inferences regarding the meaning behind the context, occasionally resulting in an understanding that differed from the respondents' intentions.