When online reviews first started in the late 1990s, shoppers used them to compare their purchases with each other instead of relying on the seller’s description or an expert recommendation. Directory sites like Yelp and Trustpilot grew their user bases by incentivizing contributors to leave authentic reviews for restaurants, doctors, and just about any public-facing business. But online reviews created a new problem: fake reviews. Businesses hired brokers to recruit people to write false reviews — either positive ones to boost their own sales or negative ones to tank sales for competitors — in exchange for cash, gift cards, or some other payment.
Now experts say artificial intelligence is adding to the problem by making it easier and faster to generate fake reviews. A recent study of 73 million reviews for home, legal, and medical services found more than 10 million were fake and more than 2 million were partly or entirely AI-generated. Last year, the Federal Trade Commission (FTC) passed a new rule to ban fake reviews written by AI as well as users who don’t have experience with a business or otherwise misrepresent their experience. The rule also bans the sale or purchase of fake reviews.
But not all AI-generated reviews are fake, which makes it difficult for the FTC to enforce its new ban. Some people use AI tools on their genuine reviews to improve grammar and word choice, and some non-native English speakers say they use AI to make sure they’re using the right words. So far, there is no clear test to differentiate between a fake review written by AI and a real review written by a person using AI tools. And research shows people are no longer able to tell the difference between reviews written by people and reviews generated by AI.
Questions:
- How can fake reviews influence shoppers who are trying to make buying decisions?
- Why did the FTC recently pass a new rule banning fake reviews, and how could the growing prominence of AI challenge this ruling?