Retrieval augmented generation
Level 2
Fetching relevant documents first, then asking the model to answer using them.
Term 5 of 7 in Data and databases
In plain language
Retrieval augmented generation, or RAG, is the common fix for hallucination and stale knowledge. Search your own material, paste the best excerpts into the prompt, and ask the model to answer only from those.
Think of it like this
An open-book exam. The knowledge is not in their head, it is on the desk in front of them.
Why it matters
RAG is how most company AI assistants actually work, and it is usually cheaper and more accurate than fine-tuning.
Next terms in Data and databases
- DatasetA collection of examples gathered for a model to learn from or be tested on.
- Structured and unstructured dataStructured data lives in neat rows and columns; unstructured data does not.
- EmbeddingA list of numbers that captures the meaning of a piece of text or an image.
- Vector databaseA database built to store embeddings and find the closest matches fast.