Zscaler Research has warned users about an illegitimate Amazon app. (Photo : Reuters)
Amazon reviews are often accused of common problems such as being one-day rush jobs, based on wrong information, or complete fakes. The online shopping giant has just launched a new machine learning system that uses artificial intelligence (AI) to sort through the reviews on Amazon, and then place the most useful ones at the top of the list, showing the most reliable, popular, and up-to-date.
The new machine-learning algorithm will impact Amazon's complex formula used to rank customer reviews and 5-star ratings system, according to Geek Wire. Amazon customers and retailers alike will be affected.
The main goal of the new software algorithm is that the reviews' quality will be greatly improved. That includes product reviews that are from verified buyers, have the most helpful votes, and include the newest updates.
In the past, Amazon would calculate ratings by using an unweighted average. The updated system has much potential, particularly for products with a high star rating, top sales figures, and many great features.
An Amazon spokesperson told GeekWire that the changes to its customer reviews system will provide more useful feedback to Internet shoppers. Its system will use machine learning to improve over time.
The new tech will roll out in the United States on Friday. Amazon users might not notice any major differences until a while later, and it is not crystal clear yet how effective the new system will be.
Ranking user reviews is not an exact science. Zero and one-star Amazon reviews will be easy to rank, but the learning technology will be unable to determine how thoughtful or earnest a product reviewer is, according to Engadget.
Then there is the issue of fake Amazon reviews. That includes high-definition TVs with the same price tag as luxury cars, or Internet cables that cause Star Trek aliens to appear.
Amazon is also currently testing a new ranking system. It will allow customers to give a number value to a certain feature of a sold product, such as reliability or battery life.
