Amazon AI: Spotting and Stopping Fake Reviews – No Fraud Allowed!
US e-commerce technology giant Amazon has launched a new and advanced artificial intelligence (AI) to identify fake product reviews on the platform. According to the company, Amazon’s AI solution analyzes the review when a customer submits a review and before it is published online for known indicators that the review is fake.
“The vast majority of reviews pass Amazon’s high bar of authenticity and are published immediately. However, if potential review abuse is detected, the company has several avenues,” the company said.
According to the company, Amazon’s machine learning models use their own data to analyze various factors. If Amazon is confident that a review is fake, it will immediately block or remove it and take additional steps, including revoking review privileges, banning fraudulent accounts, and taking legal action if necessary.
In cases where suspicion lingers but more evidence is needed, Amazon’s trained investigators, skilled at detecting abuse, look for additional signals before acting.
“Deep graph neural networks are used to explore complex relationships and patterns of behavior, helping to detect and remove groups of bad actors. Josh Meek, senior data science manager in Amazon’s Fraud Abuse and Prevention team, emphasizes that distinguishing between genuine and fake reviews is not always clear to those outside of Amazon, the e-commerce giant said.
Meanwhile, Amazon has announced it will cut hundreds of jobs from the unit responsible for popular voice assistant Alexa as the company shifts its business focus to focus on generative artificial intelligence (AI).
Alexa is an artificial intelligence-based voice assistant that can be used to set timers, make search queries, play music or as a home automation center. According to Reuters, the cuts affect several hundred employees working at Alexa.
According to the report, Amazon is streamlining the operations of various divisions this month, with cutbacks in music, gaming and certain human resources roles noted.
