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

Area of Science:

• Artificial Intelligence
• Dentistry
• Natural Language Processing

Background:

• Standalone large language models (LLMs) in dentistry face limitations such as outdated information, lack of domain specificity, and hallucinations.
• Retrieval Augmented Generation (RAG) offers a solution by combining LLMs with external knowledge bases.

Purpose of the Study:

• To review the foundational principles and architectural components of RAG.
• To explore early applications of RAG in dentistry.
• To identify challenges and future directions for RAG in dental AI.

Main Methods:

• Mini-review of existing literature on RAG and its dental applications.
• Analysis of RAG's architectural components and their integration with LLMs.
• Evaluation of RAG's impact on factual accuracy and interpretability in dental AI.

Main Results:

• RAG effectively addresses limitations of standalone LLMs, improving evidence-alignment and reliability in AI for dentistry.
• Early dental applications demonstrate enhanced factual accuracy and interpretability.
• Current research is preliminary, hindered by varied methodologies and limited dental datasets.

Conclusions:

• RAG presents a promising framework for trustworthy AI in dentistry.
• Further advancements require standardized evaluation, curated knowledge bases, and multimodal resources.
• Continued research is crucial for realizing the full potential of RAG in dental research and practice.