Research conducted by academics in Bologna has revealed that while safe shopping online can often be the best way for consumers to hunt for a bargain, the rise of AI-controlled pricing systems might make this harder to achieve in the future.

Experts warn that algorithms used to balance prices automatically across a number of major sites could ultimately move away from competitive reductions and veer towards regular increases.

By using machine learning techniques that are identical to those deployed on certain e-commerce sites, researchers were able to show that the AI could essentially work out that it is beneficial to work with rival sites and raise prices in unison. This means that without human intervention ordering products from the web might get more expensive.

This occurs because the algorithms not only learn about the ebb and flow of pricing elsewhere but also gain an understanding of how other algorithms work. If the simple goal set for algorithms to hit is an increase in profits, then it stands to reason that they might achieve this collectively through higher prices.

Most intriguing of all, this is achieved without the machine learning technologies being able to communicate directly with one another. Instead, the strategies will be developed independently and yet applied simultaneously.

Online retailers have been criticised in the past for relying on automatic pricing software which can result in some products becoming incredibly expensive without any real merit to this. Conversely, the traditional approach in the retail market has been for competing companies to cut prices in an attempt to outdo one another and earn the attention of customers.

This research will no doubt be taken into account by e-commerce sites that use machine learning to dictate pricing, since consumers will still respond negatively to inordinately expensive products.