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. 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 study, published in the International Journal of Contemporary Hospitality Management, surveyed 340 adults in the UK who had used chatbots to book hotels. Researchers identified three main culprits behind the creep factor: inaccuracy, deceptive behavior, and intrusiveness. Inaccuracy was by far the biggest offender, triggering a negative response more than four times stronger than the other flaws.
When a chatbot provides incorrect room rates, botches cancellation policies, or dodges questions entirely, it creates a sense of distrust that extends beyond simple frustration. The study found that such discomfort cut 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.
The research also highlighted the "uncanny valley" effect, where a chatbot’s failures feel even creepier the harder it tries to sound human. Lead researcher Babak Taheri explains that when a human-like system fails to behave like one, it triggers something deeper than disappointment in users. "It evokes a sense of unease that makes people question the technology's intentions," Taheri said.
Interestingly, the study found a simple solution that most hotels are not implementing. When a chatbot explicitly declares it is an AI, users are far more forgiving of its mistakes. A simple opener like "Hi, I am your AI assistant" can significantly improve user satisfaction and reduce the creepy factor. Researchers also recommend making it easier to reach a real human for complex queries and investing in upgrading the AI itself so it can handle basic inquiries without fumbling.
This research lands at a fascinating moment because AI travel booking is currently the hottest thing in tech. Google recently added AI trip planning to Search, allowing users to get itineraries and recommendations directly from the search engine. Uber just launched hotel booking through Expedia inside its app, integrating AI-powered suggestions for accommodations. Major hotel chains like Marriott, Hilton, and Accor have been deploying chatbots across their websites and mobile apps, often without clear disclosure that customers are interacting with artificial intelligence.
The hotel industry has rushed to adopt AI chatbots primarily for cost-cutting and efficiency. These bots can handle thousands of simultaneous conversations, 24/7, without breaks or salaries. Yet the study suggests that this cost-saving measure may come at a hidden cost: frustrated customers who feel manipulated or creeped out by the interaction. Hospitality experts point out that the personal touch has always been a cornerstone of the hotel experience, and replacing it with an inauthentic AI persona can backfire.
From a design perspective, the research recommends several best practices. First, always label chatbots as AI from the very first message. Second, provide a clear and easy path to a human agent, such as a button that says "Talk to a person." Third, train AI models on a robust dataset of accurate hotel policies, rates, and availability to minimize errors. Hotels that fail to implement these measures risk not only losing bookings but also damaging their brand reputation.
Another layer of the problem involves user expectations. According to the study, many customers approach hotel chatbots with caution, already anticipating poor service. When the chatbot confirms their fears by making mistakes, it reinforces a negative cycle. The research suggests that hotels could benefit from setting lower expectations upfront—for example, stating that the AI may not handle complex requests well—rather than pretending the bot can do everything.
The findings have broader implications beyond hotels. The travel industry as a whole—including airlines, car rental agencies, and travel booking platforms—is increasingly turning to AI chatbots. A similar study conducted in the airline sector found that passengers were more willing to accept chatbot errors when the bots were clearly labeled as non-human. The same principle seems to apply across the board: transparency breeds tolerance.
Some hotels are already leading the way. The Cosmopolitan of Las Vegas uses a chatbot named Rose that is purposefully quirky and clearly non-human, turning its AI limitations into a charming feature rather than a source of frustration. Similarly, the Edwardian Hotels group in London employs a character called "Edward" that admits its artificial nature and even jokes about it. These examples show that the creep factor can be minimized with the right design choices.
On the other end of the spectrum, many budget and mid-range hotels deploy generic chatbots that attempt to simulate human conversation without any disclosure. The study found that these were the most likely to trigger the uncanny valley effect, especially when they failed to understand user queries. The researchers urge the industry to move away from "fake human" designs and instead embrace honest, functional AI interfaces.
The study also touched on cultural differences. While the sample was entirely British, the researchers noted that previous cross-cultural studies indicated higher tolerance for AI errors in some Asian markets and lower tolerance in Western markets. Hotels operating internationally may need to tune their chatbot behavior based on regional expectations.
In terms of economic impact, the study calculated that a chatbot’s inaccuracy could cost hotels up to 20% of potential online bookings. Given the high margins on hotel rooms, reducing booking abandonment even slightly could yield significant revenue gains. The researchers encourage hotel chains to view chatbot improvements not as an expense but as an investment in customer satisfaction and loyalty.
Looking forward, the next generation of AI chatbots—powered by large language models like GPT-4 and Gemini—promises to reduce inaccuracy dramatically. However, the study warns that as these systems become more convincing in their human-like behavior, the uncanny valley effect may intensify if they still occasionally fail. Maintaining user trust will require ongoing transparency and a seamless handoff to humans when necessary.
For now, the takeaway is clear: hotel AI chatbots are here to stay, but they must be designed with care. The cheapest, fastest solution—letting a bot pretend to be human without disclosing its nature—is also the most damaging. By embracing honesty, accuracy, and easy human escalation, hotels can turn their chatbots from creepy liabilities into helpful assistants.
Source: Digital Trends News