ChatGPT vs. Gemini: could artificial intelligence safely improve the perioperative surgical flow while increasing patient awareness?

Abstract Aims To evaluate the efficiency and safety of ChatGPT and Gemini in the surgical workflow, from patient admission to surgical planning. Methods Twenty questions regarding preoperative and postoperative aspects of vitreoretinal surgery—most frequently asked by patients at our center—were submitted to ChatGPT (OpenAI) and Gemini (Google). Similarly, twenty surgical cases treated at our center were submitted to both platforms. The same questions and cases were evaluated by a panel of vitreoretinal surgeons with varying experience. Primary outcomes were overall reliability and inter-group agreement of the two chatbots; the secondary outcome was their ability to identify cases requiring ophthalmologist intervention. Results In the clinical case dataset ( N = 80), ChatGPT and Gemini performed comparably, with ChatGPT slightly outperforming Gemini in discussion timeliness ( p = 0.019). In the patient inquiry dataset ( N = 80), Gemini provided more complete, accurate, and timely clinical indications ( p ≤ 0.011). Stratified analyses by surgeon experience (Junior vs. Senior) showed consistent evaluations across experience levels. Conclusions ChatGPT and Gemini perform similarly in clinical case discussions, with ChatGPT slightly better in timeliness. For patient-oriented inquiries, Gemini delivers more accurate, complete, and up-to-date responses. While AI platforms demonstrate potential, their adoption in clinical practice should be preceded by thorough and rigorous evaluation.