AI & LLMs
Hallucination
When an LLM states something false with full confidence, because it predicts plausible text, not retrieves facts.
Reviewed by the RadarTrek editorial team · June 2026
A hallucination is confidently stated, plausible-sounding output that is factually wrong. It happens because a language model generates the statistically likely continuation of a prompt rather than looking facts up in a database — "plausible" and "true" are not the same thing. Rare facts hallucinate more than common ones, because the model has seen them less often during training.
Why it matters
- —Hallucination is not a bug that gets patched — it is a structural property of how generation works.
- —Grounding the model in retrieved documents (RAG) and requiring citations meaningfully reduces, but never eliminates, hallucination.
- —Safety-critical facts (medical, legal, financial) should always be verified against an authoritative source, never trusted raw.
Where to learn this
🎓
Hallucinations and Why They Happen
How LLMs Actually Work course
This is the exact lesson that covers this term in depth — with examples, diagrams, and a hands-on exercise.