Do Artificial Intelligence Chatbots Provide Adequate Information to Patients Using Implant-Supported Prostheses?

Aim: Artificial intelligence (AI) chatbots hold promise with regard to patient education because of their ability to deliver human-like responses to inquiries, yet their reliability in providing accurate information on the use and care of implant-supported prostheses – a critical aspect of prosthodontics – remains uncertain. This study sought to assess the alignment of responses from six AI chatbots to questions on this topic with the current literature on implant-supported prostheses. Materials and Method: Twenty-five questions related to the usage and maintenance of implant-supported prostheses were posed to six AI chatbots: ChatGPT-4, ChatGPT 01-Preview, ChatGPT 01-Mini, Gemini Advanced, Co-pilot, and Claude 3.5 Sonnet. The accuracy of their responses was assessed by two prosthodontists using a five-point Likert scale, and the average scores were calculated. Differences among the chatbots were analyzed using one-way ANOVA, with the significance level set at α=0.05. As the post-hoc comparison test, Tamhane’s T2 test was used. Results: The accuracy and relevance of the responses provided by the six AI chatbots to questions about the maintenance and use of implant-supported prostheses were evaluated. In terms of accuracy, ChatGPT 01-Preview achieved the highest mean score (4.80±0.08), while Co-pilot received the lowest score (3.22±0.20). ANOVA and Tamhane’s T2 tests revealed statistically significant differences between the models (p