A comparative analysis of the performance of large Language models in the dentistry specialty examination

In recent years, the rapid developments in artificial intelligence have also influenced dental education. Large Language Models (LLMs) have become increasingly accessible through publicly available chatbots and are now being used in both medical and dental training. LLMs, which have attracted significant attention across various domains, are currently being utilized in medical and dental education.The aim of this study was to evaluate the accuracy and reliability of LLM-based chatbots using questions from the Dentistry Specialization Entrance Examination (DUS), which is administered in Türkiye to assess the knowledge level of dental graduates. A total of 208 multiple-choice DUS questions were answered by seven LLMs. Data were analyzed using descriptive and comparative statistical methods, and a significance level of p < 0.05 was applied. Among the evaluated models, ChatGPT 4.0 achieved the highest accuracy (91.3%), followed by Copilot (87%) and Gemini (86.1%). ChatGPT 4.0 performed significantly better than all other LLMs (p < 0.05). In the image-based questions, the strongest performers were ChatGPT 4.0, Gemini, and Copilot, each achieving an accuracy rate of 63.6%. Although LLMs contribute substantially to dental education, their accuracy limitations in specific domains indicate that they should be used as complementary tools rather than standalone decision-makers.