Module 4 · Lesson 1 Beginner 1 min read
AI hallucinations
When AI confidently states things that are not true.
A hallucination is when AI generates false information that sounds plausible — fake citations, wrong dates, invented statistics, or non-existent products.
LLMs predict likely text; they do not verify facts. Always check important claims, especially names, numbers, legal points, and medical information.
Real examples of hallucinations
- •AI citing a paper that does not exist
- •Inventing a court case or statistic
- •Describing features a product does not have
Why hallucinations catch people out
✕ Assuming confident tone means accurate
Instead: AI can sound certain even when wrong. Verify facts.
✕ Using AI for citations without checking
Instead: Confirm papers, URLs, and statistics exist.
✕ Trusting numbers in a draft report
Instead: Cross-check figures against the source document.
Key points
- ✓Hallucinations are confident-sounding false statements
- ✓AI does not verify facts by default
- ✓Check names, numbers, and critical claims
- ✓Use AI as a draft tool, not an infallible source
Knowledge check
What is an AI hallucination?
Choose the best answer, then check your understanding.