Abstract

Background: Artificial intelligence (AI) chatbots, such as ChatGPT and Bard, have become popular sources of medical information and are likely to be used by potential kidney donors seeking information. Despite their potential role in guiding patients’ inquiries, the ability of AI chatbots to provide quality information still needs to be further investigated. This study aims to assess and compare the quality of donor nephrectomy-related information provided by ChatGPT and Bard.
Methods: A set of questions regarding kidney donation was generated based on general information from the National Kidney Foundation and the United Network for Organ Sharing. The questions were then typed directly into ChatGPT and Google Bard, and the responses were recorded and assessed for eligibility criteria. Three reviewers utilized two validated tools for evaluating health information, the DISCERN and PEMAT-P tools, to grade information quality, understandability, and actionability.
Findings: A total of 40 of 42 screened responses were included in the study, with two responses excluded for not containing information relevant to donor nephrectomies. There were no significant differences between ChatGPT and Bard based on assessment with the DISCERN, PEMAT-P Understandability, and PEMAT-P Actionability tools. Performance on the DISCERN and PEMAT-P Actionability surveys was notably poor, while most of the responses were "understandable" based on the PEMAT-P Understandability tool.
Interpretation: Both ChatGPT and Bard provide relevant and understandable responses. However, the quality of information is generally poor, and neither chatbot provides "actionable" responses. While AI chatbots have the potential for use in responding to donor nephrectomy-related queries, caution should be used.

Keywords: Donor nephrectomy; Laparoscopic; Artificial intelligence; Chatbot; ChatGPT; BARD