Glossary · Phase 1 · Building with LLMs
Hallucination
When a model states something false with confidence, because it predicts plausible text rather than checking facts.
Try it now
The documents don't answer the question. With a real model, check the reply declines instead of guessing. Then delete the sentence about saying you don't know from the system prompt, and compare.
Learn it properly
Hallucinations and grounding · Why models make things up, and how to design against it. About 25 minutes.
Other terms from the same lesson
- grounding: Making a model answer only from sources you give it, and say so when they don't cover the question.
Never trained an AI?
Teach a small one to write, right in your browser. It takes about 20 seconds and needs no account.