Intermediate · Generative AI

Grounding

Visual diagram · (in preparation) · Math · (in preparation) · Worked example · 3 difficulty levels.

TL;DR. Tying a model's outputs to verifiable, external sources of truth.

Technical Definition

Tying a model's outputs to verifiable, external sources of truth.

How it works

Grounding combats hallucination: instead of relying purely on parametric knowledge, the model is given retrieved documents, structured data, or tool outputs and instructed to base its answer on them. RAG is the most common grounding pattern; citation generation makes grounding visible to the user.

Related Concepts

• Retrieval-Augmented Generation (RAG) — A technique that enhances LLM responses by retrieving relevant documents from an external knowledge base before generating an answer.
• Hallucination — When an AI model generates confident, plausible-sounding content that is factually incorrect or fabricated.