An AI hallucination is a confident but false statement generated by a language model, such as inventing an amenity, a price or a policy the hotel does not have. It is the central reliability risk in guest-facing AI.
Hallucinations happen because LLMs generate plausible text rather than verified facts. Left ungrounded, a hotel chatbot may fabricate parking rules or breakfast hours, and an AI answer engine may describe a property inaccurately to thousands of travelers.
Mitigation is architectural. Grounding responses in verified property data, restricting what the bot may answer, and monitoring what AI systems publicly say about the property. Publishing clear machine-readable facts also reduces what external engines are left to guess.
← sesion.org · All categories · Advisors · Blog · Glossary