Purpose: Artificial Intelligence (AI) imitating human-like language, such as ChatGPT, has impacted lives throughout various multidisciplinary fields. However, despite these innovations, it is unclear how well its implementation will assist patients in clinical situations. We evaluated changes in patient perceptions regarding AI before and after reading a ChatGPT-written explanation.
Materials and methods: In total, 24 South Korean patients receiving urolithiasis treatment were surveyed through questionnaires. The ChatGPT explanatory note was provided between the first and second questionnaires, detailing lifestyle modifications for preventing urolithiasis recurrence. The study questionnaire was the Korean version of the General Attitudes toward Artificial Intelligence Scale, including positive and negative attitude items. Wilcoxon signed-rank tests were accomplished to compare questionnaire scores before and after receiving the explanatory note. A linear regression analysis with stepwise elimination was used to assess variable (demographic data) accuracy in predicting outcomes.
Results: There were significant differences between total negative questionnaire scores pre- and post-surveys of ChatGPT, but not in the positive scores. Among variables, only education level significantly influenced mean score differences in the negative questionnaires.
Conclusions: The negative perception change among urolithiasis patients after receiving the explanatory note provided by the AI chatbot program was observed, evidencing that patients with lower education levels expressed a more negative response. The explanatory note provided by the AI chatbot program could provoke an adverse change in AI perception. Negative human responses must be considered to improve and adapt new technology in health care. Only through changing patient perspectives will upgraded AI technology integrate into medical healthcare.
Keywords: Artificial intelligence; healthcare; language model; patient knowledge; ureterolithiasis.
© The Author(s) 2023.