Amazon Is Using AI To Weed Out Fake Reviews. Here’s How
Amazon recently announced it uses artificial intelligence (AI) to curb an alarming rise in fake reviews, reportedly removing 200 million in 2022. Earlier this year, the online retail behemoth reported it had taken drastic steps for what Amazon Customer Trust team head Dharmesh Mehta termed “a cottage industry of fraudsters” harming the site with fake reviews.
A Cottage Industry of Fake Reviews
Fake reviews threaten online retailers everywhere and deceive customers into buying imperfect — or sometimes fake — goods. The “cottage industry” label is apt for an industry employing thousands of people worldwide. Some customers fall for their approaches, not realizing what they are paying for.
In contrast, nefarious independent retailers can be complicit in promoting their shoddy goods for the unsuspecting public — review companies offer them thousands of reviews to boost their product sales. Moreover, some sellers offer customers gift cards or discounts for leaving five-star reviews on other products.
A High Bar for Authenticity
Amazon’s approach means customers’ legitimate reviews go through AI vetting, using a “high bar for authenticity” for immediate publishing. Meanwhile, the AI checks for potential review abuses, moving fast to block or remove the offending review and taking harsher measures when necessary, including litigation for “bad actor accounts.”
Naturally, with so many product reviews to screen and AI advancing at its current astonishing pace, AI has greatly benefited customers and businesses. Amazon’s expert investigators also search for other red flags in customers’ review activity when needed.
Poor Buying Decisions
Senior data science manager with Amazon’s Fraud Abuse Prevention team, Josh Meek, explains fake review implications in the report. “Fake reviews intentionally mislead customers by providing information that is not impartial, authentic, or intended for that product or service,” says the expert.
Millions of sellers suffer from improper review behavior, affecting customers’ judgment of their products. Moreover, customers make poor buying decisions after being convinced they have done their due diligence. Meek is adamant his team must do its utmost to protect “millions of brands and businesses.” Fake reviews also manipulate ratings so that inferior products might gain a better average rating than competitors.
Machine Learning Models
The process uses AI’s cutting-edge machine learning models to analyze large volumes of “proprietary data,” such as any seller investment in advertisements that drive extra reviews for the sellers’ wares. The AI also detects high-risk behavioral patterns, other reviews given, and prior abuse reports.
Furthermore, large language models complement natural language processing techniques, analyzing whether the review is “incentivized” with gift cards or other free offers. Amazon’s AI also operates “deep-graph neural networks” to root out relationships or other actions that might betray the reviewer.
Fake Reviews Are Hard to Spot
Meek concedes that to the non-Amazon eye, fake reviews may be hard to spot because some sellers may just be marketing a good product. “For example, a product might accumulate reviews quickly because a seller invested in advertising or offers a great product at the right price. Or, a customer may think a review is fake because it includes poor grammar.”
Amazon's AI will improve the online buying landscape if it means customers and honest sellers will have more protection. The company’s Fraud Abuse Prevention team has a powerful weapon to do the hard work thousands might otherwise have been doing.
Source: Amazon.
Raised in England and with a career background in international education, Ben now lives in Southern Spain with his wife and son, having lived on three continents, including Africa, Asia, and North America. He has worked diverse jobs ranging from traveling film projectionist to landscape gardener.
He offers a unique, well-traveled perspective on life, with several specialties related to his travels. Ben loves writing about food, music, parenting, education, culture, and film, among many other topics. His passion is Gen-X geekery, namely movies, music, and television.
He has spent the last few years building his writing portfolio, starting as a short fiction author for a Hong Kong publisher, then moving into freelance articles and features, with bylines for various online publications, such as Wealth of Geeks, Fansided, and Detour Magazine.