Influence of large language models on clinical decision-making in dentistry: A narrative review

ABSTRACT Introduction: Clinical observations, radiological interpretation, patient-specific factors, and evidence-based standards must be integrated when making clinical decisions in dentistry. Large language models (LLMs) represent a recent advancement in artificial intelligence with the potential to support complex cognitive tasks in healthcare. This narrative review aims to examine the impact of LLMs on the decision-making practices employed in the field of dentistry. Methods: A narrative review was conducted using PubMed/MEDLINE, Scopus, and Google Scholar. English-language publications between 2018 and 2025 related to LLMs, artificial intelligence, clinical decision-making, and dentistry were reviewed and qualitatively synthesized. Results: Across the 19 included articles, LLMs demonstrated evidence of supporting structured clinical reasoning and differential diagnosis generation in contexts applicable to dentistry. Some relevant findings included LLM-assisted synthesis of surgical options, identification of hallucination and automation bias as patient safety risks unique to AI-assisted surgical decision-making, and evidence from dental-specific LLM studies that accuracy in structured reasoning tasks is promising but inconsistent in complex or rare case scenarios. Patient communication, documentation, and surgical education were secondary application domains where LLMs demonstrated consistent utility across included sources. Discussion: The reviewed evidence suggests LLMs can serve as adjunctive decision-support tools in dentistry, particularly for information synthesis, structured reasoning scaffolding, and trainee education. Key gaps include the absence of prospective clinical validation studies, Specific LLM training datasets, and regulatory frameworks governing AI-assisted decisions. LLMs should supplement and not supplant the clinical judgment of the dental practitioners.