Hallucination: When information is missing, an LLM can confidently invent plausible details such as places, reviews, or facilities. It produces likely text rather than verifying that the details are true.
Overreliance: Confident, authoritative AI answers can make users overlook warning signs, especially when they lack another source of information, such as mobile reception.
Contaminated Training Data: Fictional places from novels or film scripts may appear in training data and be recommended as real attractions.