Hotel Booking Chatbots: The Creep Factor That Hurts Business

Imagine trying to book a hotel room online. An AI chatbot pops up to assist, but instead of a smooth conversation, it gives you wrong prices, misunderstands your dates, or dodges your question about cancellation policies. You feel a wave of unease. According to new research from Texas A&M University's College of Agriculture and Life Sciences, you are not alone. A study published this month reveals that hotel booking chatbots are genuinely creeping people out—and the effect is costing the industry bookings.

The researchers surveyed 340 adults in the United Kingdom who had used chatbots to book hotel stays. They identified three main factors that trigger a feeling of creepiness: inaccuracy, deceptive behavior, and intrusiveness. Of these, inaccuracy was by far the strongest offender, triggering a negative response more than four times stronger than the other two flaws combined. When a chatbot quoted incorrect room rates, botched cancellation policies, or simply failed to answer a basic question, users' discomfort spiked sharply.

The Uncanny Valley of Customer Service

The study also highlights a psychological phenomenon known as the "uncanny valley" effect. Originally coined in robotics, the uncanny valley describes the uneasy feeling people experience when something human-like—such as a chatbot or a humanoid robot—falls short of genuine human behavior. When a chatbot tries hard to sound human but then makes obvious mistakes, the failure feels especially unsettling. Lead researcher Babak Taheri explained in a statement: "When a human-like system fails to actually behave like one, it triggers something deeper than disappointment in users. It prompts discomfort, distrust, and even a sense of being manipulated."

The result is measurable. Users who encountered these creepy interactions were nearly 38% less willing to continue chatting with the bot. Moreover, the chances of delaying or completely abandoning a hotel booking nearly doubled. For hotel chains investing thousands of dollars in chatbot technology, this is a serious blow. Instead of streamlining the booking process, flawed AI is driving customers away.

The Simple Fix: Transparent Disclosure

The good news is that the researchers also discovered a straightforward remedy: transparency. When a chatbot explicitly states that it is an AI, users become far more forgiving of its mistakes. A simple opening line like "Hi, I am your AI assistant" can significantly reduce the creep factor. In the study, participants who were told upfront that they were interacting with a machine reported lower levels of discomfort and were more willing to overlook minor errors.

But many hotels ignore this fix. Instead, they deploy chatbots that mimic human names, use casual language, and avoid revealing their artificial nature. The researchers argue that this deception backfires because it raises user expectations. When the chatbot inevitably fails to meet those expectations, the letdown is amplified. "Honesty is the best policy," Taheri noted. "If you label the bot as AI, people adjust their expectations accordingly. They understand that it might not handle every nuance perfectly, and they are more patient."

The study also suggests practical steps beyond mere labeling. Hotels should make it easy to transfer a user to a human agent for complex queries. Moreover, they should invest in upgrading the AI itself so it can handle the basics—such as quoting prices, processing date changes, and explaining policies—without stumbling. A chatbot that consistently fails at these core tasks will never build trust, no matter how warmly it greets a customer.

Background: The Rise of AI in Travel Booking

This research lands at a pivotal moment. AI travel booking is arguably the hottest trend in the tech industry right now. In 2025, Google integrated AI trip planning directly into its search engine, allowing users to ask for itineraries or hotel recommendations in natural language. Meanwhile, Uber recently partnered with Expedia to enable in-app hotel bookings, further blurring the line between ride-hailing and travel services. Startups like Hopper and Kayak have also invested heavily in AI-powered chat interfaces.

However, the rush to deploy AI has often outpaced quality control. Many hotel chatbots are built on large language models that can generate convincing dialogue but are prone to hallucinations—inventing facts or rates that don't exist. In a competitive market, a single bad AI interaction can send a customer to a rival hotel that still relies on phone reservations or human-operated live chat. The Texas A&M study is a cautionary tale for the entire industry: powerful technology requires careful design and user psychology.

For context, the uncanny valley effect has been studied in robotics for decades. In 1970, Japanese roboticist Masahiro Mori first described the phenomenon, noting that as robots look and move more human, they become more appealing—but only to a point. Just before achieving full realism, there is a steep drop in likeability, a valley of eeriness. Chatbots today face a similar curve. Text that sounds human but fails to deliver accurate, context-aware responses triggers that same unsettling reaction. The study's authors believe their findings apply broadly across customer service contexts, not just hotel booking.

In addition to the three main culprits, the researchers noted that intrusiveness plays a role. Some hotel chatbots proactively ask for personal information, such as arrival time or preferred payment method, before the user has expressed a need. This can feel pushy or invasive. Others bombard users with pop-ups during the browsing experience. Combining inaccuracy with intrusive behavior creates a perfect storm of creepiness that undermines the very convenience AI promises.

To test their hypotheses, the team used a combination of survey data and controlled experiments. Participants rated their discomfort levels after interacting with different types of chatbot scripts. The results consistently showed that labeling a chatbot as AI mitigated negative reactions, while hidden or deceptive presentation made things worse. The strength of the inaccuracy effect—more than four times the impact of other factors—suggests that hotels should prioritize improving their chatbots' reliability above all else. Adding human-like qualities, like a friendly tone or a human name, can actually be counterproductive if the underlying knowledge base is shaky.

The study also examined demographic differences. While the sample was limited to UK adults, the researchers say the patterns are likely universal. People from different cultures may have different thresholds for what feels creepy, but the core emotional response to inaccurate or deceptive AI appears consistent. For global hotel chains, this means a one-size-fits-all chatbot strategy might not work across regions. Localization efforts should consider cultural sensitivities around transparency and trust.

In the broader landscape, the findings echo concerns raised by other recent studies. For instance, a 2024 report from the Pew Research Center found that half of Americans have felt uncomfortable interacting with AI customer service. Another study from MIT showed that consumers trust human agents more than AI for financial transactions. The hotel industry's reliance on chatbots is thus a double-edged sword: they save costs but risk alienating customers. The Texas A&M study offers a concrete path forward—emphasize accuracy, disclose AI identity, and always provide an escape route to a human.

For now, hotel guests can hope that this research spurs change. If you've ever felt a chill when a chatbot cheerfully gives you the wrong room price, you know exactly what the study describes. The solution, ironically, is more human: honesty and humility from the machines that try so hard to sound like us.

Source: Digital Trends News