Artificial intelligence (AI) is being used more and more to set prices online, and while this can help businesses stay competitive, it might not always be good news for shoppers.
A new study by researchers at Carnegie Mellon University reveals that personalized product rankings—when online platforms like Amazon or Expedia show you products tailored to your preferences—can actually lead to higher prices, even when sellers don’t directly target you with different prices.
AI pricing algorithms are designed to respond to market conditions and customer behavior.
While they help businesses optimize profits, they can also end up learning to keep prices higher, especially when working alongside personalized ranking systems.
The researchers wanted to find out whether these systems, which aim to help customers find the most relevant products, actually lead to better outcomes for shoppers. The answer: not always.
The team compared two types of ranking systems. One was personalized, using detailed data about each shopper to recommend products based on individual preferences.
The other was unpersonalized, showing the same product order to everyone based on average popularity or value.
Both systems help customers avoid endless scrolling by presenting a ranked list of options, but how they rank products can shape what people buy—and what they pay.
Using computer simulations, the researchers modeled how consumers search for products and how AI pricing algorithms respond. They found that in personalized systems, products predicted to suit a shopper best were shown first.
But because those top-ranked items faced less direct competition in the search list, the algorithms learned they could charge higher prices without losing sales. This led to reduced price competition and, ultimately, higher prices across the board.
In contrast, the unpersonalized ranking system encouraged more competition because no product was specially matched to a specific person. This resulted in lower prices and better overall outcomes for consumers.
The study shows that even when companies don’t engage in obvious price discrimination—charging different people different prices for the same item—personalization can still hurt shoppers by making it easier for algorithms to keep prices high. This effect remained consistent across different market conditions, learning styles, and types of products.
The findings raise important questions for tech companies, policymakers, and anyone concerned with consumer rights. If AI-powered pricing and ranking systems are not carefully regulated, they may reduce the benefits of competition and make online shopping more expensive for everyone. The study also challenges the idea that sharing more personal data with online platforms always leads to better deals. In some cases, more data can mean higher prices.
As AI continues to reshape how we shop, the way products are ranked and how algorithms set prices need careful oversight to protect consumer welfare.
