Intermediate · Evaluation

Factual Accuracy

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

TL;DR. The degree to which an AI model's generated or retrieved information aligns with verifiable facts and real-world knowledge.

Technical Definition

The degree to which an AI model's generated or retrieved information aligns with verifiable facts and real-world knowledge.

How it works

Ensuring factual accuracy is paramount for trustworthy AI, especially in sensitive domains like healthcare or news generation. Measures like RAG (Retrieval-Augmented Generation) and robust evaluation metrics are used to reduce hallucinations and improve reliability. It's a key metric for LLM performance.

Related Concepts

• Hallucination — When an AI model generates confident, plausible-sounding content that is factually incorrect or fabricated.
• Groundedness — A model's output is grounded if it directly reflects information from its provided source material.