A chatbot built on a foundation model keeps giving outdated answers because the model was trained before the company's latest documentation was written. Which technique grounds responses in the company's current documents without retraining?
A.Increase the model's temperature
Temperature controls randomness, not knowledge freshness.
B.Retrieval-Augmented Generation (RAG)
Correct. RAG grounds responses in retrieved, up-to-date documents.
C.Fine-tuning on new data
Fine-tuning works but is slower and costlier than retrieval for changing data.
D.Prompt engineering alone
Prompt engineering can't inject knowledge the model never saw.
Why this answer
RAG retrieves relevant documents at query time and injects them into the prompt — grounding the model in current data without any retraining. On AWS this is commonly implemented via knowledge bases connected to Bedrock.
Correct answer: B · Last reviewed 2026-10-26
Official AWS references
Practice this interactively
Answer without seeing the solution first in the Generative AI & Bedrock practice session.
Practice Generative AI & Bedrock