RAG (Retrieval-Augmented Generation)
Definition
A technique where an AI system retrieves relevant information from an external source and supplies it to a model when generating a response.
Put simply
The AI looks things up in information you select before answering.
What boards and leaders should understand
RAG allows organisations to combine general model capability with internal knowledge without retraining the underlying model.
Factual accuracy is not guaranteed as performance still depends on whether the right information is retrieved, whether it’s current and whether the model interprets it correctly.
Poor source information can therefore become poor AI context at scale.
Questions to consider
- How confident are we that the information being retrieved is current, relevant and appropriate for the decisions the AI is supporting?
