# Retrieval augmented generation in dentistry: potentials, applications, and future directions

## TL;DR

This paper explores how Retrieval Augmented Generation (RAG) can improve AI reliability in dentistry by combining large language models with external knowledge sources.

## Contribution

The paper introduces RAG as a novel approach to address limitations of standalone AI models in dental applications.

## Key findings

- RAG improves factual accuracy and interpretability in dental AI systems.
- Current RAG applications in dentistry are limited by heterogeneous methods and small datasets.
- Future progress requires standardized evaluation and curated dental knowledge bases.

## Abstract

Retrieval Augmented Generation offers a robust framework for developing reliable and evidence- aligned artificial intelligence in dentistry. By integrating external knowledge sources with the generative capabilities of large language models, RAG addresses key limitations of standalone LLMs, including outdated parametric knowledge, limited domain specificity, and hallucinations. This mini-review outlines the foundations of RAG, its core architectural components, and early dental applications that demonstrate improved factual accuracy and interpretability. Despite promising progress, research remains preliminary, constrained by heterogeneous methods and limited dental corpora. Advancing RAG will require standardized evaluation frameworks, curated knowledge bases, and multimodal resources to support trustworthy AI in dental research and practice.