RE: “Assessing ChatGPT's Ability to Reply to Queries Regarding Colon Cancer Screening”

Dear Editor,

We would like to comment on the published article entitled “Assessing ChatGPT's Ability to Reply to Queries Regarding Colon Cancer Screening Based on Multisociety Guidelines.¹” This research assesses the performance of ChatGPT, an OpenAI artificial intelligence-powered chatbot, in answering inquiries on colonoscopy intervals for colon cancer screening. In this study, Mukherjee et al. compared the chatbot's responses to known guidelines and evaluated its utility for patient education. The study's weaknesses include some adjudicators, which may have resulted in variability in the evaluation of the responses. Instead of actual patients, adjudicators judged the acceptability of ChatGPTTM responses for patient education. Furthermore, the chatbot's training data was only valid until September 2021, which may have impacted its accuracy. The report's future direction could include fixing the constraints by expanding the number of adjudicators and including patient feedback to evaluate the effectiveness of ChatGPTTM's responses. Updating the chatbot's training data to ensure accuracy, as well as monitoring its effectiveness in various clinical care, research, and education domains, might be examined. In conclusion, while ChatGPTTM displayed potential in answering simple questions about colonoscopy intervals, it demonstrated difficulties in handling complicated medical concerns. It has the potential to augment patient education but should not be utilized in place of information provided by a trained medical practitioner. More research and development are required to improve its accuracy and utility in clinical application.