In a simulated case study of 52 diverse clinical scenarios, investigators compared two large language model -based chatbots - ChatGPT (GPT‑4) and Gemini (1.5 Flash)—on their ability to deliver guideline‑concordant, personalized dietary advice for diabetes. Using an Analytic Hierarchy Process framework, Gemini achieved higher overall scores for clinical relevance, evidence alignment, personalization, and practicality, with particularly better performance in complex situations such as diabetes with chronic kidney disease, malnutrition, or post‑bariatric surgery.
While both tools showed promise, the authors emphasize that current AI systems should augment - not replace - clinical judgment, and call for ongoing refinement, validation, and governance before routine integration into diabetes nutrition counseling. To read more click here ##Reference## Zhu Y, Li X, Wang J, et al. Assessing the…
