Glossary
What is AI hallucination?
AI hallucination is a confident answer from an AI model that is wrong or invented, with no sign of doubt.
Why it happens
A model writes the most likely next words, not the most verified facts. When it lacks an answer, it may fill the gap with plausible detail. It sounds sure of itself, so the mistake is easy to miss.
Where it hurts most
Prices, policies, delivery dates, medical or legal details and product specifications carry the highest risk. A wrong answer can cost a sale or create a complaint. Customers act on what a bot tells them.
How to reduce it
Give the model trusted source material, and tell it to answer only from that. Let it say it does not know, and hand the chat to a person. Test with real customer questions before launch, and review transcripts every week.
Example
A Dubai travel agency’s chatbot is asked about visa fees. Without a source, the bot quotes an old fee. With the agency’s current visa page as its only source, the bot gives the right fee. Any question outside that page goes to an agent.
Montr’s website chatbot is still in beta. It answers only from the documents and forms you connect, and it hands over to your team when it cannot answer.