Confiabilidade e completude das respostas de inteligências: artificiais generativas (chatgpt e gemini) em odontopediatria: comparação com evidências científicas da pubmed – estudo piloto

Abstract

INTRODUCTION: The advancement of digital technologies, especially Artificial Intelligence (AI), has impacted Dentistry. Tools such as ChatGPT and GEMINI are widely used by students and professionals for study and clinical support. However, research evaluating whether their answers align with current scientific literature is still scarce. OBJECTIVES: To analyze whether the answers from ChatGPT (paid and free versions) and Gemini (paid and free) on topics in Pediatric Dentistry correspond to the

knowledge present in systematic reviews from PubMed. METHODOLOGY: This cross- sectional, descriptive, and analytical study evaluated the reliability and completeness

of the answers provided by four artificial intelligence platforms (free ChatGPT, paid ChatGPT, free Gemini, and paid Gemini) in comparison with high-level scientific evidence. Five relevant clinical questions in Pediatric Dentistry were selected and answered using systematic reviews and guidelines identified through a structured search in the PubMed database, constituting the gold standard. The same questions were submitted, in a standardized way, to the AI platforms. The responses were analyzed for reliability and completeness. RESULTS: The four AI platforms showed varying performance. The paid version of ChatGPT demonstrated the highest reliability, completeness of answers, conceptual accuracy, and best alignment with scientific evidence when compared to the paid version of Gemini and the free versions. The free versions, however, showed a greater tendency towards generic, superficial answers with less technical rigor, insufficient for academic or clinical applications that demand methodological precision. FINAL CONSIDERATIONS: Although the paid version of ChatGPT stood out compared to the other models, none of the AIs evaluated achieved sufficient consistency to replace systematic reviews or clinical guidelines as primary sources of scientific information.