Tripadvisor's artificial intelligence-generated summaries of hotel reviews are systematically downplaying serious complaints, including those about safety and hygiene, according to a new study published in the Journal of Consumer Research.
Researchers from the University of California, Berkeley, and the University of Chicago analyzed over 100,000 hotel reviews on Tripadvisor and found that the AI summaries, which are designed to provide a quick overview of guest feedback, tend to omit or soften negative comments that are critical of the hotel's safety standards, cleanliness, or health hazards.
Study findings: AI summaries miss critical feedback
The study, led by Professor David Smith, examined reviews for hotels in major US cities and compared the full text of reviews with the AI-generated summaries. The researchers found that the summaries were 40% less likely to mention issues related to safety, such as broken locks, mold, or pest infestations, compared to the original reviews. Similarly, complaints about cleanliness were 35% less likely to appear in the summaries.
“The AI system appears to be programmed to prioritize positive or neutral language, which can mislead consumers who rely on these summaries for quick decision-making,” said Smith. “Important warnings about health and safety are being filtered out, which could have serious consequences for travelers.”
How the AI summaries work
Tripadvisor introduced AI-generated summaries in 2024 as a feature to help users quickly grasp the general sentiment of reviews. The system uses natural language processing to extract key themes and condense multiple reviews into a short paragraph. However, the study suggests that the algorithm may be biased toward highlighting common positive phrases while downplaying rare but critical negative comments.
The researchers noted that the summaries often use vague language like “some guests had concerns” or “a few issues were noted” without specifying the nature of the problems. In contrast, positive comments are frequently quoted directly, such as “great location” or “friendly staff.”
Impact on consumers and hotels
The findings raise concerns about consumer protection in the digital age. Many travelers use review summaries to decide where to stay, especially when browsing on mobile devices. If serious complaints are omitted, they may unknowingly book hotels with safety risks.
“This is a classic case of algorithmic bias,” said co-author Dr. Emily Johnson. “The AI is optimizing for readability and positive sentiment, but at the expense of accuracy and completeness. Consumers deserve to see the full picture, including the bad reviews.”
The study also found that hotels with lower overall ratings were more likely to have their negative comments downplayed, potentially masking patterns of poor service or unsafe conditions.
Tripadvisor responds
In response to the study, a Tripadvisor spokesperson said the company is committed to transparency and is continuously improving its AI systems. “We take feedback seriously and are reviewing our algorithms to ensure they provide balanced and accurate summaries,” the spokesperson said. “Our goal is to help travelers make informed decisions, and we will work to address any issues identified by this research.”
The company noted that users can always read the full reviews by clicking on the summary, and that the AI feature is optional. However, critics argue that many users rely solely on the summary, especially on mobile devices where scrolling is limited.
Calls for regulation
The study adds to growing calls for regulation of AI-generated content in consumer-facing platforms. Consumer advocacy groups have urged platforms like Tripadvisor to be more transparent about how their algorithms work and to include disclaimers about potential biases.
“This is not just about convenience; it’s about safety,” said Mark Taylor, director of the Consumer Rights Coalition. “When AI systems downplay serious complaints, they are effectively misleading consumers. We need rules that require these summaries to be accurate and complete.”
The researchers recommend that Tripadvisor and similar platforms should allow users to customize the summaries to highlight critical issues, such as safety or hygiene, and should provide a clear option to see the full set of reviews without any AI filtering.
Broader implications for AI in reviews
The issue extends beyond hotel reviews. Similar AI summarization tools are used on platforms like Amazon, Yelp, and Google Maps. The study suggests that the same biases could be present in those systems, potentially affecting consumer choices across multiple industries.
“Our findings have implications for any platform that uses AI to summarize user-generated content,” said Professor Smith. “As AI becomes more common in e-commerce and travel, it is crucial that these systems are designed to prioritize accuracy and transparency over brevity and positivity.”
