Mining vast troves of data, companies like Delta and Instacart are using AI to set prices for everything from airline tickets to groceries. As the AI gets smarter, it may figure out just what you'd be willing to pay. By Callum Keown
Consumers are using ChatGPT and other artificial-intelligence apps to book vacations and shop online. What they may not realize: Companies are using AI to squeeze more from your wallet.
Mining vast troves of data, AI is setting prices for everything from airline tickets to groceries. The tools are getting smarter as they learn our shopping habits. They're also getting personal, generating prices and offers that hinge partly on what the AI thinks you're worth as a customer and would pay.
How this works varies by industry, but it's becoming pervasive. Delta Air Lines has started using AI to set fares for some domestic tickets. Retailers such as Home Depot are using AI to analyze data on consumers and then tailor prices based on the AI's knowledge.
Royal Caribbean CEO Jason Liberty said recently that AI is managing "15 million price points" a day, displacing human revenue managers. "Our AI tools are getting smarter and smarter so that we're able to curate what is relevant to that consumer," he said on an earnings call in October.
Consumers are in the dark about AI's machinations. Companies are loath to discuss details of their pricing algorithms. Partly that's for competitive reasons. A consumer and regulatory backlash is also brewing over AI being used for unfair trade practices and illegal forms of price discrimination.
The technology is certainly powerful: AI has the potential to turbocharge "personalized pricing," a method of charging different prices to different sets of consumers or individuals. Personalized pricing isn't new. We already pay different prices for many things -- airline tickets, home insurance, groceries -- based on our consumer data or ability to haggle in real life (in the case of a car, for instance).
AI's twist is that it is constantly getting smarter. The computing is powerful enough -- thanks largely to Nvidia's chips -- that AI can blend our data with the dynamics of a marketplace, tailoring prices to individuals or groups more narrowly. AI is also a high-tech crystal ball, capable of guessing what you might be willing to pay at a point in time, and customizing an offer to you.
"Consumer data isn't just fueling personalization; it's redefining pricing," says IDC research manager Adam Wright. "With AI tools mining billions of data points in real time, prices can now better reflect demand, loyalty, and context, which creates a new frontier in dynamic pricing."
If AI helped everyone gets better deals -- while pumping up corporate profits -- there wouldn't be much concern. But academic studies indicate that AI-powered algorithms, working competitively, wind up raising prices for everyone. In 2019, academic researchers in Europe concluded that AI algorithms "consistently learn to charge supra-competitive prices," resulting in tacit collusion. More recent research has found similar findings.
Some Democratic politicians see AI as a potential tool for "surveillance pricing," the idea that companies are closely tracking consumers and charging more to some individuals or groups based on personal data. The Biden administration started investigating the practice, though the Trump administration dropped the probe. Under Republican control, the federal government appears unlikely to set rules for AI-based pricing.
Some states are trying to put up guardrails. New York state recently passed a law requiring companies to disclose if they're using algorithmic pricing. California state lawmakers have proposed regulations, though they didn't make it to a vote. The Trump administration, meanwhile, is trying to stop such moves, recently directing the Department of Justice to sue states that issue AI-related regulations.
Publicly, companies are walking a tightrope. While they're eager to show investors that they're making good use of AI, they risk antagonizing consumers and lawmakers if it appears that they're using the software for unfair or discriminatory trade.
Delta drew wrath from Democratic senators after the company said it was expanding its use of AI for pricing. Delta subsequently said it isn't using AI to set personalized ticket prices.
"I think what's most important is that it's done in a noncreepy way," said Royal Caribbean's Liberty at a recent conference, adding that the company's AI tools aren't being used to monitor customers' habits but instead to predict what they might like.
Yet experts see a disconnect between how companies say they're using AI -- fairly for all -- and its potential to charge some consumers more than others.
"The way this is being pitched to investors is that, through hyperpersonalization, companies will be able to significantly raise revenue and margins -- that they'll be able to charge the maximum each consumer is willing to pay," says Samuel Levine, former director of the Federal Trade Commission's Bureau of Consumer Protection in the Biden administration.
Airlines, Amazon, and AI
Airlines are integrating AI tools, aiming to improve operations and bolster revenue with more-personalized prices and offers.
Delta is partnering with an Israeli software company called Fetcherr that has created 4-D models of the market -- combining supply/demand information with consumer and corporate data. The models can adjust and set prices far faster and more accurately than older software. Fetcherr says its "large market model" can uncover "hidden" revenue, boosting sales by an average 10% while cutting manual processes by 60% compared with traditional software.
Wall Street loves that idea. AI should "drive [a] pricing tailwind in 2026," said TD Cowen analyst Tom Fitzgerald in a recent note.
Much of the software is still in training mode, but it's likely to get more powerful as it incorporates more data and learns from its mistakes. Delta initially had a target of using AI to price 20% of its domestic tickets by year end, but the figure is now likely to be closer to 5%, according to a person familiar with the matter.
Whether consumers see higher or lower prices is debatable. But some research indicates that the algorithms can push through higher prices.
Today, products you see on Amazon.com, Expedia, TikTok, and other sites are ranked not just by how useful they are to you, but also by prices pegged to rival sellers in the marketplace. Knowing more about your preferences, AI may recognize that you're less price sensitive and show you higher-priced goods first. You may see lower-priced goods ranked higher if you're profiled as more price sensitive.
That might seem to level the playing field in aggregate. But a recent study by Carnegie Mellon University business professor Param Singh and other researchers concluded that personalized rankings, using AI-algorithms, increase prices overall for consumers by an average 29%, or 13% after accounting for the economic benefits of more convenience.
"When algorithms personalize how products are ranked for each user, they tend to learn to charge higher prices overall," Singh says.
AI chatbots also appear effective at boosting revenue. Customers using Amazon's AI-powered Rufus tool are 60% more likely to make a purchase, said CEO Andy Jassy on the company's earnings call at the end of October. Rufus is on track to drive an extra $10 billion in annual sales for the company, he said.
Even if you don't buy something with Rufus, the chatbot can use the context it gains about you from chats for a period of 28 days (assuming you don't opt out), according to Amazon's privacy policy. The privacy notice says it uses your personal information to recommend products and services.
OpenAI's ChatGPT is moving into e-commerce, too. Companies such as Walmart, Etsy, and Shopify have launched partnerships with ChatGPT to capture sales through the app. The chatbot can use information from your queries to personalize your shopping.
OpenAI said in a statement to Barron's that ChatGPT's results are selected independently and aren't influenced by ads or its partnerships. If a consumer specifies a budget, the chatbot will focus on that, but if price isn't mentioned, it may home in on other factors.
Behind the scenes, AI is learning far more about how consumers shop. Home Depot partners with AI-pricing firm Revionics, for instance. The tools could be used to track and analyze a contractor's recent purchases of building materials, for example, and provide a bundled discount for the next phase of the job. Home Depot's privacy policy says consumers may see a "different rate" based on the value of their personal information.
Revionics told Barron's that its software doesn't use individual consumer data but considers factors such as historical sales data and competitive prices. It said its clients "leverage science to find prices that will win the loyalty of their target customers," adding that it often recommends price cuts.
Consumers may also be guinea pigs as companies use AI and other software tools to experiment with prices. And discounts may be far less prevalent than markups.
Instacart, for one, recently charged up to 23% more to different consumers for the same grocery items at the same store, according to an investigation by Consumer Reports and Groundwork Collaborative. Almost three-quarters of grocery items were offered at multiple price points, the report found. A dozen Lucerne eggs at a Safeway store in Washington state went from $3.99 to $4.79, for instance.
While consumers might not notice a few pennies here and there, it can add up. The report found that the same basket of groceries at a Seattle-area Safeway on Instacart ranged from $114.34 to $123.93. Only 8% of shoppers got the best deal, and the cost swings over a year could add up to $1,200 for a family of four, the report said.
(MORE TO FOLLOW) Dow Jones Newswires
December 19, 2025 21:30 ET (02:30 GMT)
Copyright (c) 2025 Dow Jones & Company, Inc.
Comments