Answer

RAG: How AI Looks Things Up Before Answering

AI models only know what they learned during training, which stops at a certain date. RAG, short for Retrieval-Augmented Generation, lets an AI pause before answering, search through a pile of real documents, grab the relevant bits, and then use those bits to write its answer. Think of it like an open-book test instead of a closed-book one. The AI still does the writing, but it can peek at fresh, accurate information first. This makes answers more reliable and up-to-date without retraining the whole model.

Example: A customer service chatbot uses RAG to search your company's latest policy documents before answering a refund question, so it never gives outdated advice.