Can Artificial Intelligence Chatbots (ChatGPT, DeepSeek, Copilot, and Claude) Accurately Inform Patients About Subperiostal Jaw Implants?

Abstract

Background: Severe jawbone atrophy often limits the use of conventional dental implants. Modern 3D imaging and CAD/ CAM technologies have revitalized subperiosteal implants as a customized alternative for these challenging cases. As AI chatbots (ChatGPT, Claude, DeepSeek, and Copilot) increasingly serve as patient information tools, evaluating the accuracy and clarity of their explanations about such complex procedures has become essential. Objective: This study aimed to evaluate the accuracy, reliability, readability, actionability, understandability, and practical usefulness of responses provided by AI chatbots to patient questions about subperiosteal jaw implants. Methods: The authors evaluated 4 AI-based chatbots (ChatGPT, DeepSeek, Copilot, and Claude) by submitting frequently asked questions on subperiosteal jaw implants in independent sessions to avoid data leakage. Responses were compiled, duplicates removed, and refined for clarity. Two independent experts assessed the outputs using validated tools: accuracy (5-point Likert), reliability (CLEAR criteria), quality (mGQS), readability (FRE, FKGL), usefulness (4-point scale), and understandability/actionability (PEMAT). Results: DeepSeek, Claude, and ChatGPT produced more understandable, actionable, and higher-quality responses than CoPilot, with DeepSeek performing the best overall. Across all models, clarity, mGQS, and accuracy were strongly aligned, while usefulness was inversely related. Readability and actionability-understandability correlations showed consistent patterns, with the strongest positive link observed in DeepSeek. Conclusion: AI chatbots, such as DeepSeek, Claude, and ChatGPT, can provide accurate and understandable in formation about the subperiosteal jaw implants, though practical guidance and readability remain limited. Domain-specific training and integration with authoritative dental resources may enhance their clinical utility and patient education potential.