AI summaries of Tripadvisor hotel reviews downplay serious complaints, investigation finds

A hotel being sued for mass food poisonings was described as “spotless” and a resort where guests complained of sexual harassment by staff was praised for “friendly” service by an AI intended to summarise millions of Tripadvisor reviews.The overviews of customer feedback downplayed serious complaints, ranging from the stench of mould to a lack of mains water, according to an investigation by the consumer campaign organisation Which?The AI-generated reviews appear on the travel website’s hotel webpages to help holidaymakers decide where to book. In one case, the AI summary described the Riu Palace Santa Maria in Cape Verde as popular, with spacious rooms, “diverse restaurants” that earn “rave reviews” and spotless cleanliness.Yet customers reported being served raw chicken, shared photographs of flies and birds on the buffet and reported “dead little roasted mice by the sitting area”. A guest whose whole family became ill wrote: “This place will destroy holidays.”The hotel chain is being sued in the high court by hundreds of guests alleging illnesses linked to poor hygiene standards and food safety failings.The AI-generated review for the Cape Verde hotel is no longer available. Its operator, RIU Hotels & Resorts, said: “We operate with the highest standards of professionalism and service, placing hygienic-sanitary safety as our top priority.”Tripadvisor said it was monitoring and refining its AI tool but it was ‘confident these features are delivering exactly what they were designed to do’. Photograph: Jakub Porzycki/NurPhoto/Rex/ShutterstockThe AI praised a separate Dominican Republic hotel for its “abundant” amenities, with only a nod to “inconsistent” cleanliness and “maintenance issues”. But guests reported showering with bottled water because the mains taps ran dry and that every other person in a large wedding party became sick.Meanwhile, guests of a hotel in Turkey wrote that they felt unsafe due to repeated sexual harassment from male hotel staff, including requests to connect on social media, but the AI summary described the service as “friendly” and merely said “lapses [in service] noted by a few”, Which? found.Tripadvisor said it was monitoring and refining its AI tool and was “looking into the examples where reviews did not match the intended property”. But it said it was “confident these 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”.It said the AI-generated summaries did not replace travellers’ reviews and that customers had the common sense to check any AI advice against the billion-plus reviews and contributions it has gathered.Rory Boland, the 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.”skip past newsletter promotionafter newsletter promotionTripadvisor said its systems automatically suppress AI summaries when travellers warn about serious safety incidents such as death, drugging or sexual assault, “helping ensure this content is highly visible to our community”.Duncan Brumby, a professor of human-computer interaction at University College London, said the case chimed with his own research into academics using AI during the peer-review process. He found AI tends to “sanitise and rub off the edges” of sharper criticisms, likely because the bulk of the training data contains so many more bland observations.“Here you have guests describing a really negative experience, but the AI has decided to tone it down,” he said. “It’s as if it’s being polite.”Other studies have found technologies designed to summarise opinions often suffer from a blind spot that means they reduce the richness of consumer feedback to shallower sentiments.This year Google removed some of its AI health summaries after a Guardian investigation found people were being put at risk of harm by false and misleading information.