Investigation: Tripadvisor AI summaries downplay serious hotel complaints
Consumer group Which? found that AI-generated summaries on Tripadvisor often gloss over critical issues like food poisoning and sexual harassment, misleading travelers.
An investigation by the consumer campaign organization Which? has revealed that Tripadvisor's artificial intelligence (AI) generated hotel review summaries frequently minimize or misrepresent serious complaints.
According to the investigation, a hotel currently being sued for mass food poisoning was described in its AI summary as “spotless”. Another resort where guests reported sexual harassment by staff was praised for “friendly” service. Other examples include complaints about mold smell and lack of running water being downplayed.
Specifically, the Riu Palace Santa Maria in Cape Verde, which faces a high court lawsuit from hundreds of guests over hygiene failures, was praised by the AI for “diverse restaurants” and “spotless cleanliness”. However, guests reported raw chicken, flies and birds on the buffet, and even “dead little roasted mice by the sitting area”. One guest wrote: “This place will destroy holidays.” The operator, RIU Hotels & Resorts, said it maintains high professional and hygiene standards.
At a separate hotel in the Dominican Republic, guests had to shower with bottled water because the mains ran dry, and many wedding party members fell ill, yet the AI summary only mentioned “inconsistent” cleanliness.
In Turkey, guests felt unsafe due to repeated sexual harassment from male staff, including requests for social media contacts, but the AI summary described service as “friendly” and merely noted “lapses by a few”.
Tripadvisor said it is monitoring and refining its AI tool and investigating the specific examples. The company emphasized that AI summaries do not replace user reviews and that travelers have common sense to verify information. However, Rory Boland, editor of Which? Travel, advises users to scroll past these summaries and read original reviews, especially one-star ratings.
Professor Duncan Brumby from University College London noted that AI tends to “sanitize and rub off the edges” of sharp criticisms because training data contains many bland observations. Other studies show that summarization tools often reduce the richness of consumer feedback to shallower sentiments.
