How is artificial intelligence implemented in dentistry, and what are its applications, performance, and benefits compared to traditional methods in recent studies?

Background: Artificial intelligence has been progressively incorporated into dentistry, particularly in diagnostic imaging, treatment planning, predictive modeling and patient education. The expansion of machine learning, deep learning and language-based models has enabled new approaches to clinical decision support and workflow optimization.Aim: To synthesize the available evidence on the implementation of artificial intelligence in dentistry, focusing on its clinical applications, performance, comparison with traditional methods and reported limitations.Methods: A systematic review was conducted that included studies published between 2020 and 2025. A comprehensive search was performed in PubMed, Scopus, Web of Science, LILACS and ProQuest. The selection process included identification, duplicate removal, screening by title and abstract, and full-text assessment. Studies were included if they reported applications of artificial intelligence in clinical, diagnostic or educational contexts within dentistry. Data extraction considered the study design, application area, type of artificial intelligence, purpose, performance metrics, comparison with conventional methods and main findings. The evidence was synthesized through a structured narrative approach because of methodological heterogeneity.Results: Artificial intelligence is widely applied in radiological diagnosis, orthodontics, implantology, periodontics and educational settings. Deep learning models, particularly convolutional neural networks, are predominant in image-based tasks and demonstrate high diagnostic accuracy. In structured clinical scenarios, artificial intelligence systems achieve performance comparable to or exceeding that of traditional methods. Improvements are observed in diagnostic precision, standardization of procedures and decision support. Performance variability is identified in complex clinical tasks, where contextual interpretation remains necessary. Most studies present limitations related to sample size, lack of external validation and heterogeneity in methodological approaches.Conclusions: Artificial intelligence represents a relevant tool for supporting clinical and educational processes in dentistry, particularly in structured and data-driven tasks. Its implementation shows consistent benefits in terms of accuracy and standardization. Current evidence indicates that these systems should complement, rather than replace, clinical judgment. Further research with a rigorous design and external validation is needed to determine the generalizability and clinical impact of these findings.