AI-generated summaries of Tripadvisor hotel reviews have been found to downplay serious complaints, including allegations of sexual harassment and food poisoning, according to an investigation by consumer campaign organization Which?.
AI summaries gloss over serious issues
The AI overviews, designed to help holidaymakers quickly gauge hotel quality, described a hotel being sued for mass food poisonings as “spotless” and a resort where guests complained of sexual harassment as having “friendly” service. In one case, the Riu Palace Santa Maria in Cape Verde was praised for its “spotless” cleanliness, yet guests reported raw chicken, flies on buffets, and “dead little roasted mice by the sitting area.” The hotel chain is being sued in the high court by hundreds of guests citing poor hygiene.
Rory Boland, editor of Which? Travel, said: “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.”
Specific examples of misleading summaries
At a Dominican Republic hotel, the AI praised “abundant” amenities with only a nod to “inconsistent” cleanliness, but guests reported showering with bottled water because mains taps ran dry, and many in a wedding party became ill. In Turkey, an AI summary described service as “friendly” despite guests feeling unsafe due to repeated sexual harassment from staff, including requests for social media contact.
Duncan Brumby, professor of human-computer interaction at University College London, said AI tends to “sanitise and rub off the edges” of sharp criticisms, likely because training data contains many bland observations. “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,” he said.
Tripadvisor response and broader context
Tripadvisor stated it is monitoring and refining its AI tool and is “looking into the examples where reviews did not match the intended property.” It expressed confidence that the features “are delivering exactly what they were designed to do: help travellers quickly understand the breadth of feedback while making it easy to explore the underlying reviews in full.” The platform noted that AI summaries are automatically suppressed when travellers warn about serious safety incidents like death, drugging, or sexual assault.
Other studies have found that AI summarization technologies often reduce the richness of consumer feedback to shallower sentiments. This mirrors issues seen with Google’s AI health summaries, which were partially removed after a Guardian investigation found they could cause harm with false information.
