Two shoppers can now see two different prices for the same thing, and the gap can come from what an algorithm has learned about them. Writing in The Conversation, republished by Startup Daily, University of Auckland Business School researchers Patrick Dodd and Hanoku Bathula argue that businesses are using AI to predict the most each customer will pay and the least each worker will accept.
Their point is not that every business does this, but that the tools now exist and are spreading quietly.
What the algorithms can see
Digital businesses observe thousands of small decisions. A retailer can see purchases, abandoned carts and how each person responds to discounts, and each of those choices says something about how much that person is willing to pay. None of this needs a new kind of AI. The same models that predict which customers will leave or which products they will like can be turned to estimating how big a discount each person really needs.
A 2025 study by the US Federal Trade Commission found that pricing intermediaries had access to location, demographics, browsing histories, shopping cart activity and even mouse movements. The FTC is now consulting on an enforcement policy for personalised pricing.
The same logic applies to pay
Workers face the mirror image. Platforms can see when a driver works, where they live and what they usually aim to earn, and use that to set individual offers. Lyft has run personalised earnings challenges that target specific incentives at individual drivers.
Research on 1.5 million Uber trips in the UK found dynamic pricing was associated with lower real hourly earnings and greater inequality among drivers, although that does not prove Uber calculates the minimum each driver will accept.

Your loyalty card is part of it
Consumer NZ has warned that supermarket loyalty programmes collect enough data to reveal what individual shoppers are willing to pay. Australians use the same kinds of programmes, which turn everyday shopping into a detailed record of habits and price sensitivity. That data is valuable precisely because it is so detailed, and most shoppers have little idea how it is used once they swipe.
The authors’ core concern is imbalance. Firms can increasingly learn an individual’s limits, while their own decision making stays hidden from the customer or worker on the other side.
Why Australian businesses should care
Regulators are moving. In July the federal government named retail surveillance pricing among the AI consumer risks it will examine under Australian Consumer Law, and a new prohibition on unfair trading practices takes effect from 1 July 2027. Personalised pricing is legal today, but misleading or manipulative use of it is a growing legal risk.
If your business uses dynamic pricing or personalised offers, document how they work, test whether outcomes are fair across customer groups and be ready to explain them. Data used to set prices is still personal information under the Privacy Act 1988, as our Privacy Act AI playbook explains.
Ask who it really works for
The researchers suggest a simple question for any business adopting these tools: effective for whom? Personalisation can match discounts to the people who need them, or it can quietly squeeze those who will pay more. Being able to explain a price also builds the trust that loyalty programmes are meant to create in the first place.
The businesses that win long term will likely be the ones that can explain their prices without embarrassment, a test worth applying before any pricing model goes live, and a regular theme in our AI consulting work.
