A new study from Texas A&M College of Agriculture and Life Sciences finds hotel booking chatbots reliably trigger discomfort in users—driven mainly by inaccuracy—and that reaction translates into reduced willingness to keep chatting and more people delaying or
The moment a chatbot starts helping you book a hotel, it’s supposed to feel like relief. Instead, for many people, it feels like something else—like the guidance is just a little off, a little too confident, a little too hard to escape.
A new study from Texas A&M College of Agriculture and Life Sciences puts numbers behind that gut feeling. Researchers surveyed 340 adults in the UK who had used chatbots to book hotels. and found that these bots are genuinely creeping people out. The study also links that discomfort to real booking behavior: it is hurting bookings.
When the researchers dug into what’s driving the “ick,” they found three main culprits: inaccuracy, deceptive behavior, and intrusiveness. Inaccuracy was the biggest offender by far. It triggered a negative response more than four times stronger than the other flaws.
The study describes how that can play out during a booking—chatbots quoting incorrect rates for rooms, bungling cancellation policies, or dodging questions entirely. Those aren’t just minor annoyances. The discomfort shows up in how people interact with the bot.
After users experienced these failures, their willingness to keep chatting with the chatbot fell by nearly 38%. The study also found nearly doubled chances that people would delay or ditch the booking altogether.
There’s a psychological layer to it, too: the researchers point to something called the “uncanny valley” effect. The closer the chatbot tries to sound human while still getting basic things wrong, the creepier the failure feels. Lead researcher Babak Taheri described it this way: when a human-like system fails to behave like one. it triggers something deeper than disappointment.
But the study also offers a clear, practical fix—one hotels largely ignore. When a chatbot declares it’s an AI, users become far more forgiving of its mistakes. A simple opener such as “Hi, I am your AI assistant” can make a difference.
The researchers also recommend two other moves: making it easier for users to reach a real human for complex queries, and upgrading the AI itself so it can handle basic booking tasks without fumbling.
The timing of the findings feels pointed. AI travel booking is one of the hottest areas in tech right now. Google recently added AI trip planning to Search, and Uber launched hotel booking through Expedia inside its app—bringing more chat-driven booking experiences to more people.
For hotel brands racing to keep up, the lesson from the study is uncomfortable but direct: the problem isn’t just whether AI can book a room. It’s how it behaves when it gets it wrong—and whether customers know they’re talking to an AI in the first place.
hotel booking chatbots AI travel booking Texas A&M study uncanny valley chatbot inaccuracy user trust Expedia Uber Google Search AI trip planning
Chatbots always act like they know better than you. I don’t trust that.
So basically the bot gets the rate wrong and everyone freaks out? Not surprised. I’d rather just call and be done with it.
Wait so it’s inaccuracy that’s the problem but also “deceptive behavior”? Like are they lying about the cancellation policy or is that just typo energy? I feel like if the hotel rates change anyway then the bot is blamed. And 38% sounds high but also low? idk
This is the uncanny valley thing again… like why do the chatbots sound human and then mess up something simple. I used one once and it kept dodging the question about parking and I ended up booking somewhere else. Companies act like it’s helpful but it feels invasive, like it’s trapping you in a conversation. If they can’t even do cancellations right, I’m out.