Retrieval Augmented Generation (RAG) is a technique that enhances the capabilities of generative models by integrating external knowledge sources. It works by retrieving relevant documents or information from a database or a search engine, and then feeding this information into a generative model like a language model. This enables the system to generate more informed and accurate responses, enriching the context that the model operates within. Compared to traditional generative models, RAG reduces the risk of hallucination by grounding outputs in real-world data, similar to techniques like open-domain question answering systems.