AWS Certified AI Practitioner AIF-C01 Practice Question

Retrieval Augmented Generation (RAG) links a foundation model with an external knowledge base. When a user submits a question, relevant passages are first retrieved from the company's documents (stored, for example, in S3 and indexed in a vector database). These passages are then provided as additional context to the model, allowing it to generate an answer that is grounded in the source material. This lowers hallucination risk and avoids the cost and effort of full fine-tuning. Simply adjusting inference parameters or performing parameter-efficient fine-tuning would not connect the model to the manuals, and RLHF is a training-time technique unsuited to dynamically incorporating up-to-date documents.