Definition
An AI hallucination is a confident but false or unsupported statement produced by an AI model, such as an invented policy, price, feature, or citation.
Language models generate text that is statistically plausible, not text that is checked against facts. When the model lacks the right information, it may fill the gap with something that sounds right.
In customer support, hallucinations are costly: a bot that promises a refund the policy does not allow, or a feature the product does not have, creates real obligations and angry customers.
Mitigations include grounding answers in approved content (RAG), instructing the model to say it does not know, limiting the topics it may answer, showing source links, reviewing transcripts, and handing off to a human when confidence is low.
Example
A customer asks a bot about a student discount. With no article on the topic, the bot invents "15% off with a student ID", a discount the company has never offered.
How Yaplet helps
Yaplet's AI agent answers from your own website and knowledge base content and can hand the conversation to your team when it cannot answer.