Key facts
- Tripadvisor's AI summaries of hotel reviews have been found to downplay serious complaints, according to a Which? investigation.
- Serious issues such as mass food poisoning allegations and sexual harassment by staff were described with neutral or positive language by the AI.
- One AI summary described a hotel facing a lawsuit for food poisoning as "spotless" and having "diverse restaurants" with "rave reviews".
- Another AI summary characterized sexual harassment complaints at a Turkish hotel as "friendly" service with "lapses noted by a few".
- Tripadvisor stated it is monitoring and refining its AI tool and is confident in its current functionality.
An investigation by the consumer group Which? has revealed that AI-generated summaries of hotel reviews on Tripadvisor frequently downplay serious guest complaints. The AI tool, designed to help travelers quickly understand feedback, presented a hotel facing a lawsuit for mass food poisoning as "spotless" with "diverse restaurants" that receive "rave reviews." In another instance, a hotel where guests reported sexual harassment by staff was described by the AI as having "friendly" service, with only minor "lapses" noted.
These summaries have faced criticism for sanitizing negative experiences, potentially misleading travelers. Guests at the Riu Palace Santa Maria in Cape Verde, for example, reported encountering flies and mice in food areas and experiencing severe illness, with one stating, "This place will destroy holidays." The hotel chain, RIU Hotels & Resorts, maintains it operates with the highest standards of safety.
Similarly, a Dominican Republic resort praised by the AI for "abundant" amenities and "inconsistent" cleanliness was also reported by guests to have issues with dry taps and widespread illness among a wedding party. The AI's description of a Turkish hotel's service as "friendly" contrasted sharply with guests' accounts of feeling unsafe due to repeated sexual harassment from staff.
Tripadvisor acknowledged the findings, stating it is monitoring and refining its AI tool, and is "looking into the examples where reviews did not match the intended property." The company expressed confidence that the features are functioning as designed, helping users explore underlying reviews. Rory Boland, editor of Which? Travel, urged users to bypass AI summaries and consult detailed guest reviews, especially one-star ratings, and check other sites for safety information.
Experts suggest AI's tendency to smooth out sharp criticisms may stem from training data containing more bland observations. Professor Duncan Brumby of University College London noted that AI might "tone down" negative experiences, likening it to being "polite." Previous instances of AI providing misleading health information, such as Google's removed AI health summaries, highlight the risks associated with AI summarization technologies.