Lexicon / CONCEPT
Hallucination
When an AI model generates information that sounds plausible but is factually incorrect, fabricated, or not grounded in its training data.
A hallucination occurs when an AI model confidently produces incorrect or entirely made-up information. The output reads naturally, which makes it especially tricky to spot.
Why It Happens
- Models predict the most likely next token — they don’t “know” facts
- Higher temperature settings increase creative (and sometimes wrong) outputs
- Questions outside the training data push the model to guess
How to Reduce Hallucinations
- Use RAG to ground responses in verified sources
- Lower the temperature for factual tasks
- Ask the model to cite sources or say “I don’t know”