When a model produces fluent but unsupported or incorrect claims.
How to Debug With AI Assistants
How to Debug With AI Assistants — operator steps you can finish in one working session.
AI Glossary
Hallucination
When a model produces fluent but unsupported or incorrect claims.
Simple explanation
A hallucination is when an AI answer sounds confident but is wrong or unsupported.
Technical explanation
Models can generate fluent but ungrounded claims; mitigation includes retrieval, citations, constrained decoding, and human review.
Examples
- Invented citations
- Wrong API parameters stated confidently
FAQs
How do I reduce hallucinations?
Ground answers in sources, ask for uncertainty, and review high-stakes outputs.
Related terms
Retrieval-Augmented Generation (RAG)
A pattern that grounds model answers in retrieved documents or knowledge bases.
ExploreAffiliate Disclosure
Affiliate Disclosure is a working definition teams use while practicing How to Evaluate AI Tools.
ExploreAgency Knowledge Base
Agency Knowledge Base is a working definition teams use while practicing AI for Marketing Agencies Playbook.
Explore