Artificial intelligence in dentistry: clinical applications, ethical challenges and future perspectives

Introduction: Artificial intelligence (AI) has progressively transformed healthcare across multiple specialties. In dentistry, deep learning algorithms and convolutional neural networks have demonstrated diagnostic accuracy comparable to, and in certain contexts superior to, that of experienced clinicians. This narrative review synthesizes the current evidence on AI applications in dentistry, with particular attention to the ethical challenges and specific barriers faced in Latin American contexts, including Bolivia. Methods: A narrative review of the scientific literature was conducted using PubMed, Scopus, and Google Scholar. Articles published between 2025 and 2026 were included, prioritizing systematic reviews, meta-analyses, and peer-reviewed original studies. The following MeSH terms were used: "artificial intelligence," "dentistry," "ethics," "machine learning," and "oral diagnosis." Development: The first diagnostic support systems were explored in the 1980s and 1990s. The qualitative leap came with deep learning and CNNs, which allow for the analysis of radiographic images with a granularity that, in many contexts, surpasses inter-observer variability. CNN-based systems have demonstrated consistent results in caries detection, periodontal evaluation, identification of periapical lesions, and diagnosis of maxillary pathologies in panoramic, periapical, and cone-beam computed tomography (CBCT) radiographs. AI has generated particularly robust results in automated cephalometric analysis. Conclusions: AI in dentistry has reached clinical maturity in diagnosis, showing accuracy comparable to or better than that of human experts, especially in dental radiology. Its implementation in Latin America and Bolivia is conditioned by structural inequalities, such as low investment, a lack of local data, and limited digitization, which could widen health gaps if not addressed adequately. The future of dentistry will depend not only on how much AI is used, but also on how well technology is balanced with the professional's responsibility, ethics, and clinical judgment.