Evaluation & safety · Glossary term

What is Hallucination?

Generated content that is false, unsupported by the available evidence, or inconsistent with the task's source of truth. It can arise even when the output is fluent and the model is not attempting to deceive.

What people say

“The model is lying.”

Why does Hallucination matter?

You usually cannot inspect whether a statement existed in training data, so production checks should focus on support, correctness, and traceability.

Hallucination in practice

Require cited evidence for factual answers and evaluate whether each citation actually supports the associated claim.

What is the common confusion about Hallucination?

A hallucination is an output-quality failure, not a diagnosis of model intent.

Learn Hallucination in the course

Lessons that name Hallucination in a title or section

  • Failure Modes: Why Agents Break

    MASFT (Berkeley, 2025) catalogs 14 multi-agent failure modes in 3 categories. Microsoft's Taxonomy documents how existing AI failures amplify in agentic settings.

    Phase 14: Agent Engineering

Covered in Phase 14: Agent Engineering.

  • GroundingConnecting a generated answer or action to evidence, state, or observations that the system can identify and check.
  • RAG (Retrieval-Augmented Generation)A system pattern that retrieves evidence relevant to a request and supplies selected content to a generative model before it answers or…
  • Verification GateA control point that blocks progress until defined evidence satisfies a correctness or quality criterion.

More terms in Evaluation & safety

Open the Evaluation & safety list in the glossary

This entry comes from glossary/terms.md on GitHub. Browse all 250 glossary terms.