AI Summaries Understate Serious Hotel Issues
An AI-generated overview intended to summarize millions of Tripadvisor hotel reviews has been found to downplay serious complaints, including allegations of sexual harassment and hygiene problems, according to an investigation by the consumer campaign organisation Which?.
One example involved a hotel being sued for mass food poisonings, which the AI described as “spotless.” Another resort, where guests reported sexual harassment by staff, was praised for “friendly” service.
Cases of Misleading AI Summaries
The AI-generated reviews appear on Tripadvisor’s hotel webpages to assist holidaymakers in making booking decisions. In one instance, the Riu Palace Santa Maria in Cape Verde was described by the AI as popular, featuring spacious rooms, “diverse restaurants” earning “rave reviews,” and spotless cleanliness.
However, customers reported raw chicken being served, shared photographs showing flies and birds on the buffet, and noted “dead little roasted mice by the sitting area.” One guest, whose entire family became ill, wrote:
“This place will destroy holidays.”
The hotel chain is currently facing a high court lawsuit from hundreds of guests alleging illnesses linked to poor hygiene standards and food safety failures.
The AI-generated review for the Cape Verde hotel has since been removed. RIU Hotels & Resorts, the operator, stated:
“We operate with the highest standards of professionalism and service, placing hygienic-sanitary safety as our top priority.”
Additional Examples of AI Downplaying Issues
Another hotel in the Dominican Republic received AI praise for its “abundant” amenities, with only a brief mention of “inconsistent” cleanliness and “maintenance issues.” Guests, however, reported having to shower with bottled water due to mains water shortages and noted that many members of a large wedding party fell ill.
Meanwhile, guests at a hotel in Turkey reported feeling unsafe due to repeated sexual harassment by male hotel staff, including requests to connect on social media. Despite these serious allegations, the AI summary described the service as “friendly” and only mentioned “lapses [in service] noted by a few,” according to Which?.
Tripadvisor’s Response and Expert Opinions
Tripadvisor acknowledged the findings and stated it was monitoring and refining its AI tool, adding it was “looking into the examples where reviews did not match the intended property.” The company expressed confidence that its AI features are delivering their intended purpose:
“help travellers quickly understand the breadth of feedback while making it easy to explore the underlying reviews in full.”
Tripadvisor emphasized that AI-generated summaries do not replace traveller reviews and that customers typically verify AI advice against the extensive collection of over a billion reviews and contributions.
Rory Boland, editor of Which? Travel, commented on the issue:
“The platform has a responsibility to revisit the accuracy of its AI summaries and AI chatbot. In the meantime, users should scroll past these summaries and look at guest reviews, particularly one-star ratings, and at reviews on other sites, to make sure their next stay is a safe one.”
Tripadvisor also stated that its systems automatically suppress AI summaries when travellers report serious safety incidents such as death, drugging, or sexual assault, ensuring such content remains highly visible to the community.
Duncan Brumby, professor of human-computer interaction at University College London, related the findings to his own research on AI use in academic peer review. He observed that AI tends to “sanitise and rub off the edges” of sharper criticisms, likely because training data predominantly contains more neutral observations. He said:
“Here you have guests describing a really negative experience, but the AI has decided to tone it down. It’s as if it’s being polite.”
Other technologies designed to summarize opinions often face similar challenges, reducing the complexity of consumer feedback to simpler sentiments.
This year, investigations have revealed that false and misleading information generated by AI can put people at risk of harm.
