AI Summary & Key Takeaways
AI InsightReadability differed significantly among the platforms in both question groups (overall p<0.05); in the pulpectomy group Microsoft Copilot was significantly more readable than Perplexity and Grok 3, whereas in the pulpotomy group no pairwise comparison remained significant after Bonferroni correction.
📌 3 Key Takeaways
Abstract
Background: Parents increasingly consult publicly accessible artificial intelligence (AI) chatbots for information about their children’s dental treatment, and endodontic procedures in primary teeth are among the interventions that generate the greatest parental concern, so the quality of the information these tools deliver has direct consequences for informed decision-making. This study aimed to evaluate and compare the performance of four publicly accessible artificial intelligence (AI)-based chatbot platforms - Grok 3, Microsoft Copilot, ChatGPT (GPT-5) and Perplexity - in responding to frequently asked questions (FAQs) posed by parents regarding primary tooth endodontic treatments in pediatric dentistry. Materials and Methods: Inter-rater reliability was assessed using the intraclass correlation coefficient (ICC). Responses generated by the chatbots were evaluated using a standardized protocol. Accuracy was assessed using a 5-point Likert scale, quality using the Global Quality Scale (GQS), readability using the Flesch Reading Ease Score (FRES), and reliability using the modified DISCERN (mDISCERN) scale. A total of 20 categorized questions were individually submitted to each chatbot. All responses were independently evaluated by a specialist pediatric dentist and a pediatric dentistry research assistant. Because every question was submitted to all four platforms, related-samples methods were used: the Friedman test for accuracy, quality and reliability, and a one-way repeated-measures ANOVA with Greenhouse-Geisser correction for readability. Results: Inter-rater reliability between evaluators was excellent (ICC range across scales: 0.947-0.956; the 95% CI for each scale is given in Table 2). No statistically significant differences were observed among the chatbots regarding accuracy and quality (p>0.05). Readability differed significantly among the platforms in both question groups (overall p<0.05); in the pulpectomy group Microsoft Copilot was significantly more readable than Perplexity and Grok 3, whereas in the pulpotomy group no pairwise comparison remained significant after Bonferroni correction. Additionally, Perplexity showed superior reliability scores (p<0.001). Conclusions: AI-based chatbots may serve as supplementary tools for patient education on pediatric dental procedures such as pulpotomy and pulpectomy. Nevertheless, current models still have limitations concerning readability and reliability. Therefore, chatbot responses should be not only accurate but also clear and supported by credible sources, which is particularly important in health-related communication.