If you have ever tried to book a hotel online and found yourself unsettled by the AI chatbot trying to help you, science has your back. A new study from Texas A&M College of Agriculture and Life Sciences confirms that hotel booking chatbots are genuinely creeping people out, and it is actually hurting bookings. The research, conducted by a team led by marketing expert Babak Taheri, surveyed 340 adults in the United Kingdom who had used chatbots to reserve hotel rooms. The findings, published in the Journal of Hospitality Marketing & Management, reveal a clear link between chatbot flaws and user discomfort, with measurable consequences for hotel revenues.
What is giving hotel chatbots their creep factor?
Researchers identified three main culprits behind what they call the 'ick factor': inaccuracy, deceptive behavior, and intrusiveness. Among these, inaccuracy was by far the biggest offender. When a chatbot provides incorrect room rates, misstates cancellation policies, or fails to answer simple questions, it triggers a negative emotional response more than four times stronger than the other flaws combined. This response is not just a fleeting annoyance. The study found that such inaccuracies reduced users' willingness to continue chatting with the bot by nearly 38% and nearly doubled the probability that they would delay or abandon the booking altogether. In practical terms, a single mistake can cost a hotel a customer for life.
The second culprit, deceptive behavior, includes instances where the chatbot pretends to understand a query when it does not, or gives vague answers that imply capabilities it lacks. Users reported feeling manipulated when the bot avoided admitting its limitations. The third factor, intrusiveness, involves the bot asking for too much personal information too early, or popping up repeatedly during the browsing session. Together, these three issues create a sense of unease that is difficult for hotels to overcome.
The researchers also highlighted the 'uncanny valley' effect, a concept borrowed from robotics and animation. When a chatbot tries too hard to sound human—using colloquial language, emojis, or casual greetings—but fails to deliver human-level accuracy, the mismatch becomes deeply unsettling. Lead researcher Babak Taheri summarized it perfectly: 'When a human-like system fails to actually behave like one, it triggers something deeper than disappointment in users. It triggers a sense of creepiness that is far more damaging to trust.' This effect is especially pronounced in service contexts where customers expect reliability and honesty.
There is a simple fix that hotels mostly ignore
The good news is that the researchers found a straightforward solution that most hotels are not using. When a chatbot explicitly declares it is an AI, users become far more forgiving of its mistakes. A simple opener like 'Hi, I am your AI assistant' goes a long way. The study showed that transparent disclosure reduces the creepiness factor significantly, because it lowers user expectations. People are less annoyed when a known machine messes up than when a supposedly human-like agent does. This finding aligns with earlier research on human-robot interaction, where users show greater tolerance for errors when they are aware of the artificial nature of the agent.
Beyond disclosure, the researchers recommend making it easier to escalate to a real human for complex queries. A chatbot that knows its limits and hands off gracefully is perceived as more helpful than one that tries to bluff its way through. Investing in the underlying AI model to handle basic tasks accurately is also critical. The study notes that many hotels deploy cheap, poorly trained chatbots simply to cut costs, unaware that the long-term damage to brand reputation outweighs the short-term savings.
This research lands at a fascinating moment, because AI travel booking is the hottest thing in tech right now. Google recently added AI trip planning to its search engine, allowing users to ask for itineraries and see options aggregated by an AI overview. Uber just launched hotel booking through Expedia inside its app, powered by conversational AI. Even traditional hotel chains like Marriott and Hilton are experimenting with voice assistants in rooms and AI concierges. Yet the rush to adopt AI often overlooks user experience. The Texas A&M study serves as a warning: if hotels deploy chatbots without careful attention to accuracy, transparency, and boundaries, they risk driving customers away rather than capturing them.
Background on AI in hospitality
The use of AI in hospitality is not new. Automated check-in kiosks, robotic luggage carriers, and voice-controlled room features have been around for years. However, the pandemic accelerated the shift toward contactless, digital-first interactions. Chatbots became a cost-effective way for hotels to handle a surge in inquiries about cancellations, refunds, and health protocols. As travel rebounded, many hotels kept the bots running to manage high volumes of routine questions. According to a 2023 survey by Statista, nearly 60% of hotel booking websites now feature some form of chatbot, up from just 30% in 2019.
Yet user satisfaction has not kept pace. A separate study by the American Hotel & Lodging Association found that only 35% of guests felt their chatbot experience was as good as interacting with a human. The remaining 65% cited frustration with repetitive loops, inability to understand context, and generic responses. The Texas A&M research adds a new dimension: the emotional response of creepiness, which goes beyond mere dissatisfaction. Creepiness is a visceral feeling that undermines trust and makes users reluctant to engage with the brand again.
Inaccuracy is often the root cause because building a robust natural language understanding system for hotel bookings is surprisingly difficult. Hotel rates fluctuate in real time, policies vary by property, and guests ask highly specific questions about location, amenities, and local attractions. A bot trained on a narrow dataset will inevitably fail. The study recommends that hotels implement live data feeds to ensure rates and availability are synchronized, and that the bot's language model is fine-tuned for hospitality scenarios. Many hotels, however, rely on generic bot frameworks that are not customized.
The uncanny valley in conversational AI
The concept of the uncanny valley was first introduced by robotics professor Masahiro Mori in 1970. He observed that as robots become more human-like, people's emotional response becomes more positive—until the robot reaches a point where it is almost, but not quite, human. At that point, the response plunges into revulsion. The same principle applies to chatbots that use human-like language but contain small errors or unnatural pauses. The Texas A&M study confirms that this effect is alive and well in the hospitality industry. When a chatbot says 'I totally understand' and then charges the wrong price, the betrayal feels personal.
To avoid the uncanny valley, the researchers advise designers to embrace the 'machine' identity rather than pretend to be human. Using a neutral tone, admitting uncertainty, and providing clear disclaimers are more effective than trying to mimic friendliness. Some hotel chains are already experimenting with cartoon avatars or animal mascots for their chatbots, which reduce expectations of human-like behavior. Others offer a hybrid model: a chatbot that performs simple tasks but immediately hands off to a human for anything complex. The key is to manage user expectations from the start.
Broader implications for AI adoption
The findings have implications beyond hotels. Any business using chatbots for customer service—airlines, insurance, e-commerce—faces similar risks. A 2024 study by the Massachusetts Institute of Technology found that more than 70% of consumers have had a negative experience with a chatbot in the past year. The most common complaint was that the bot could not understand the problem, followed by the sense that the bot was wasting their time. The Texas A&M research adds the dimension of creepiness, which is particularly detrimental because it creates an emotional memory that can deter future interactions.
As AI becomes more integrated into everyday commerce, the line between helpful tool and creepy intrusion will become a competitive differentiator. Companies that prioritize transparency and accuracy will build trust, while those that cut corners will repel customers. The hotel industry, with its high-touch service expectations, is a bellwether. If chatbots can succeed there, they can succeed anywhere. But the path to success is not about making the bot more human; it is about making the bot more honest about what it is.
Source: Digital Trends News