Two freshly delivered Amazon boxes are seen on a counter in Golden, Colorado August 27, 2014. Reuters/Rick Wilking has deployed a new artificial intelligence learning system to filter out fake product reviews. The system, built by the retailer, aims to improve customer service by emphasizing newer reviews.

E-commerce sellers and advertisers commonly post fake product reviews on Amazon, Yelp, Trip Advisor and other sites where customer reviews are essential to a buyer's decision. Amazon's new AI tries to curb this problem, known as a astroturfing, by giving more weight to reviews from verified customers, newer reviews and reviews that have been marked as “helpful” by other Amazon customers. It will also be capable of filtering product listings that have been modified to address customer complaints.

“The system will learn what reviews are most helpful to customers...and it improves over time,” Amazon spokeswoman Julie Law told CNet. “It's all meant to make customer reviews more people see things and know it reflects the current product experience.”

The program was put into place Friday, though how long it has been under development wasn't immediately clear.

The change coincides with Amazon's announcement that it will begin testing a program that pays eBook authors based on the number of pages customers read. Starting July 1 writers who self-publish through Amazon's KDP Select Program will earn royalties whenever a reader stays on an eBook page long enough for that page to be read. The system is widely expected to help authors who include cliffhangers and plot twists in their story as well as benefit the writers of long books, who have complained that Kindle versions of their work sell for the same price as a slim paperback.