Assessing the Accuracy and Reliability of Artificial Intelligence in First Permanent Molar Extraction Decisions

Abstract

The first permanent molar (M1) plays a critical role in maintaining functional occlusion during mixed dentition. When this tooth is severely compromised by deep caries, structural weakness, or developmental enamel defects, extraction may be required. However, the decision to extract M1 is influenced by multiple prognostic factors, including the child’s age, dental development stage, second molar (M2) angulation, and the presence of third molars. In this context, artificial intelligence (AI)–based conversational systems can support clinical judgment and patient education by offering quicker access to relevant information. This descriptive-analytical study evaluated the performance of two AI-powered language models (ChatGPT-3.5 and ChatGPT-4) in providing clinical guidance on M1 extraction. Ten standardized questions were formulated based on established prognostic parameters associated with spontaneous space closure. Both AI systems produced coherent and clinically relevant content. However, ChatGPT-4 tended to produce longer responses with greater sentence complexity, resulting in lower readability and higher academic demand. By comparison, ChatGPT-3.5 produced shorter, more straightforward responses that may be more appropriate for communication with parents or patients. Future research incorporating varied question structures, such as case-based scenarios, radiographic-based questions, and real-user feedback, may help define how conversational AI can be safely and effectively integrated into pediatric dental workflows. Until then, AI-based tools should be used as supportive resources rather than as independent clinical decision-makers, and their outputs should be reviewed by qualified professionals to ensure patient safety.