Back to Glossary

Hallucination

NLP & Language Models

Plausible but incorrect or fabricated model output.


Hallucination refers to the tendency of a model to produce false or fabricated information that sounds convincing.

  • Causes: Lack of context, outdated knowledge, probabilistic text generation.
  • Risks: Misinformation, loss of trust, ethical and legal issues.
  • Mitigation: Retrieval grounding, source citation, and fact-checking layers.