Travelers who use chatbots powered by artificial intelligence on hotel booking platforms often feel uneasy. That discomfort can cause them to disengage or delay booking decisions, according to new research.
The study, from the Texas A&M College of Agriculture and Life Sciences and published in the International Journal of Hospitality Management, found consumers react most negatively to hospitality chatbots when the technology appears inaccurate or unreliable, deceptive or intrusive.
Researchers also identified a simple way companies can improve customer trust: clear AI disclosure and transparency.
“Chatbots are rapidly becoming a critical element of hospitality service,” said Babak Taheri, lead researcher and professor in the Arch H. Aplin III ’80 Department of Hospitality, Hotel Management and Tourism. “But when chatbots try to sound human and fall short by giving inaccurate prices, dodging questions or behaving inconsistently, users don’t just get annoyed — they get creeped out, and they leave.”
People are also reading…
The study’s authors determined “creepiness” by examining how outdated, inappropriate, incomplete or inadequate information affects conversational accuracy and triggers unsettling feelings in users while interacting with hospitality chatbots.
Inaccuracy triggers the strongest reaction
Researchers analyzed survey responses from 340 adults in the United Kingdom who used chatbots to book hotels.
Applying the Stimulus-Organism-Response framework, a model commonly used in consumer psychology, they examined how chatbot shortcomings create creepy feelings and shape user behavior.
All three chatbot flaws increased user discomfort, but inaccuracy produced the strongest negative response, with a path coefficient more than four times larger than incredibility.
The study found that the discomfort reduced users’ willingness to continue interacting with the chatbot by nearly 38% and nearly doubled the likelihood they would delay or abandon a booking decision. A chatbot that provides incorrect room rates or misinterprets a cancellation policy can drive customers to another platform.
Researchers also identified the “uncanny valley” effect, in which chatbot failures become more unsettling as the technology more closely mimics human speech.
“Users implicitly expect a human-like system to behave like a human,” Taheri said. “When it doesn’t, the gap between expectation and reality triggers something deeper than disappointment. It triggers a threat response.”
Simple disclosure can reduce creepiness
The study also examined whether users respond differently when a chatbot identifies itself as AI at the start of a conversation.
On one hand, researchers found this disclosure intensified the creepiness feeling linked to incredibility because users became more sensitive to deception once they knew they were interacting with a machine.
However, the disclosure reduced discomfort tied to inaccurate responses. Researchers said users were more likely to attribute mistakes to AI limitations rather than deliberate deception.
The researchers recommend hotel platforms begin chatbot conversations with a clear statement such as, “Hi, I’m your AI assistant,” and provide an easy way for users to connect with a human representative for more complex questions.
They also urged hotel and travel companies to upgrade the AI technology behind their chatbots, so these virtual assistants can better manage important customer questions about room availability, prices and cancellation policies.
Running experiments that compare different approaches side by side was another recommendation, aimed at finding the clearest, most effective way to communicate key information to customers and the best time to share it.
“Competence without transparency doesn’t solve the problem,” Taheri said. “Hospitality companies need both if they want customers to trust AI-powered service interactions.”
