Evaluation of Artificial Intelligence Responses on Molar Incisor Hypomineralization

Abstract

Molar–incisor hypomineralization (MIH) is a developmental enamel defect involving one or more first permanent molars and often the incisors. Affected molars are structurally fragile, making them highly susceptible to post-eruptive breakdown and dental caries. As the condition often causes hypersensitivity, pain, and esthetic concerns, many parents and patients increasingly seek information from online platforms. In this context, artificial intelligence (AI)–based chatbots have emerged as accessible tools for delivering health-related information. This study aimed to evaluate and compare the performance of six AI-driven chatbots—ChatGPT-3.5, ChatGPT-4, Gemini Advanced, Microsoft Copilot, Claude 3, and Perplexity AI—in providing educational and clinical information about MIH. Twelve standardized questions derived from the 2021 European Academy of Paediatric Dentistry (EAPD) guidelines were presented to each chatbot. Their responses were analyzed using EQIP, DISCERN, Global Quality Score (GQS), Flesch Reading Ease Score (FRES), Flesch–Kincaid Grade Level (FKGL), and Similarity Index. All chatbots generated coherent and clinically relevant answers. ChatGPT-4 achieved the highest EQIP (64.10) and reliability (3.70) scores, followed by ChatGPT-3.5 and Gemini Advanced. ChatGPT-3.5 recorded the highest GQS (4.60), while Gemini exhibited the lowest Similarity Index (1.50), indicating more unique phrasing. All outputs showed college-level readability (FRES < 50; FKGL 10.3–12.2). These findings suggest that OpenAI-based models perform best in generating reliable and informative MIH-related content, although their high readability level may limit accessibility for the general population.