Amazon reviews became longer and less helpful almost overnight in December 2022 when shoppers started using AI to describe products they hadn't used.
April 23, 2026
Original Paper
When AI Writes the Reviews: Generative AI and the Changing Nature of User-Generated Content in Online Markets
SSRN · 6529658
The Takeaway
Generative AI has caused a structural break in the quality of user-generated content on major retail platforms. Reviews in categories where experience can be easily faked are now homogenized and filled with plausible-sounding filler. People used to rely on these reviews as a layer of trust based on genuine human experience. Now, the influx of synthetic text is polluting that trust and making it harder for consumers to make informed choices. This trend indicates that the era of the reliable online review may be coming to an end.
From the abstract
The launch of ChatGPT in November 2022 reduced the cost of producing written content to near zero, with consequences for online review ecosystems that remain poorly understood. Using an interrupted time series design on nearly 50 million Amazon reviews across ten categories, we find that reviews became longer, progressively less helpful, and more similar to one another within category-month bins. An endogenous structural break test places the helpfulness break date exactly on December 2022 witho