(P16-020-25) Comparing Registered Dietitian and AI Chatbot Responses to Nutrition-Related Questions on a Public Social Media Forum

Objectives: Artificial intelligence (AI) has been incorporated into nutrition counseling rapidly, with chatbots such as ChatGPT offering accessible, personalized dietary advice. An evaluation of the impact of AI and RD responses on patient care is proposed in this study, which is the first systematic comparison of their responses in nutrition counseling.

Objective: This study aims to examine AI-generated and RD responses to nutrition-related questions, focusing on clinical accuracy, empathy, readability, and response length.

Methods: There will be a total of 100 questions randomly chosen from public social media forums.Responses generated by AI and RDs will be evaluated by highly experienced clinical dietitians using a scoring framework. Clinical accuracy and empathy will be assessed using 5-point Likert scales (1 = poor, 5 = excellent). Readability will be measured using the Flesch-Kincaid Grade Level, and response length will be recorded in word count. The overall response quality score (0–100) will incorporate assessments of clinical accuracy, empathy, readability, and response length. Data analysis will include t-tests to compare group means, correlation analyses to examine relationships between variables, and sensitivity assessments to identify the impact of response length on perceived quality and empathy.

Results: This study is expected to provide insight into the strengths and limitations of AI nutrition counseling and RD nutrition counseling, including their ability to provide evidence-based nutrition counseling. Based on the literature review, we hypothesize that the RD responses will demonstrate strengths in empathy and nuanced clinical decision-making, while the AI responses will demonstrate strengths in readability and response length.

Conclusions: This study will highlight key gaps in AI-driven nutrition counseling, specifically in emotional engagement and sophisticated clinical judgment. The purpose of these findings is to underscore how combining AI technologies with human knowledge can improve patient outcomes and increase access to evidence-based nutrition treatment.

Funding Sources: It did not receive any funding from organizations, institutions, or agencies.