AI-generated summaries of product reviews significantly increase consumer purchase intent, according to a University of California, San Diego study. Participants were 32% more likely to want to buy a product after reading an AI summary than a human-written review.
The research analyzed 1,000 electronics reviews, 1,000 media interviews, and 8,500 news items using six large language models. The AI hallucinated in 60% of cases when asked about factual details. It also altered the sentiment of real user reviews in 26.5% of instances.
Lead researcher Abeer Alessa found that LLMs frequently misrepresent events post-training and suffer from 'lost in the middle' bias-focusing on early text while ignoring later information.
Even in low-stakes scenarios, the findings raise red flags for high-risk domains like healthcare documentation and college admissions, where biased framing could distort judgment with real-world consequences.