During Amazon’s semiannual Prime Day bonanza, the underwear I usually buy were on deep discount at a price so low I assumed it was just for me, the frequent shopper. I got three packs and felt good about it. But a few days ago, when I saw the news that Delta Air Lines was using AI to personalize its flight prices, I got mad. I started to suspect that something sinister was behind my great deal on drawers at Amazon. Was that sale just for me? Where else have I been paying personalized prices?
Delta CEO Glen Hauenstein actually announced last year that the airline was using AI to do “a full reengineering of how we price and how we will be pricing in the future” at its annual Investor Day event. He promised, somewhat ominously, that airfares would be determined “on that flight, on that time, to you, the individual.” Last week, Hauenstein ignited a furor telling investors that Delta was currently using the technology on 3 percent of flight prices and planned to boost that to 20 percent by the end of this year.
Welcome to the era of hyper-personalized pricing. Companies are increasingly deploying AI-powered technology that is capable of identifying thousands of different real-time signals — everything from your location and loyalty status to your device and search history — to sell the same product to two different people for two different prices. This represents an advanced form of dynamic pricing, the age-old practice of adjusting prices based on market conditions. With the help of algorithms and reams of data, some businesses are taking a new, personalized approach: surveillance pricing. Dynamic pricing is perfectly legal, but surveillance pricing and the accompanying privacy concerns are new.
Suffice to say, consumers don’t like the idea of companies using AI to set prices. On Wednesday, Democratic Rep. Greg Casar announced plans to introduce a bill that would ban surveillance pricing at a federal level. Arizona Sen. Ruben Gallego accused Delta of “using AI to find your pain point — meaning they’ll squeeze you for every penny” and sent the company an angry letter, which was cosigned by Sens. Mark Warner of Virginia and Richard Blumenthal of Connecticut. And earlier this month, New York enacted a law that requires sellers to disclose when personalized algorithmic pricing is in effect, with similar legislation under consideration in other states.
Consumers don’t like the idea of companies using AI to set prices.
When I asked George Slover, general counsel of the Center for Democracy and Technology, about pricing models like this, he said, “This is a different animal than what the airlines have been doing in the past, and it is more personalized and more intrusive.”
Delta denies anything untoward is happening here. The company said in a statement, “There is no fare product Delta has ever used, is testing, or plans to use that targets customers with individualized offers based on personal information” and that fares are “based solely on trip-related factors like advance purchase and cabin class.” In other words, according to the company, AI helps Delta set prices using metrics the airline already uses to determine airfares. Amazon, too, says it doesn’t use surveillance pricing after a botched experiment 25 years ago.
Still, surveillance pricing is already a documented phenomenon. Kroger offers different discounts to different customers based on personal data, according to a Consumer Reports investigation published in May. Target settled a lawsuit and paid $5 million in fines after a local KARE 11 news investigation in Minneapolis found that prices in the Target app changed when customers entered the perimeter of a store. And a ProPublica investigation revealed that the Princeton Review charged Asian families higher prices for college prep services.
“It is a more sophisticated and algorithmically driven and selective price gouging,” Slover said of surveillance pricing, which he calls bespoke pricing. “You are focusing on one particular individual based on their vulnerability and susceptibility.
Dynamic pricing — sometimes with an algorithmic boost — is also controversial and on the rise. In the past year alone we’ve seen outrage over irregularities in ticket prices for Beyoncé’s Cowboy Carter tour, an AI-enabled menu at Wendy’s that offered items at different prices depending on the time of day, and rent-setting software from RealPage that the Department of Justice say enabled collusion between landlords. And going further back, you’ve got Uber allowing massive surge pricing during Hurricane Sandy and Amazon doing random price tests on customers buying DVDs.
If you’re wondering who to blame for this trend for this era of computer-optimized price schemes, the answer is surprisingly obvious: It’s airlines. Well, and Jimmy Carter.
Airline prices have always been a black box
The era of dynamic pricing as we know it started in 1966, when American Airlines launched its Semi-Automated Business Research Environment, or SABRE. This computerized reservations system became the company’s nerve center, where data about every reservation and cancellation was kept.
After President Carter signed the Airline Deregulation Act into law in 1978, which allowed airlines to set their own prices, American Airlines leveraged the data they collected through SABRE to maximize profits. The company even created a new system called DINAMO to do so, and within a few years, dynamic pricing became industry standard. American Airlines led the way, launching its Super Saver fares and helping drive down the price of flights for leisure travelers while business travelers footed the bill for the discounts. However, it wasn’t always clear why some seats were cheaper than others. This is still the black box approach to airfare pricing we’re living with today.
Airfares can vary widely based on a number of factors, from the number of discounts the airline releases to when the passenger books the flight. The person you’re sitting next to on any given flight could have paid twice as much as you — or half as much. It’s getting harder to figure out how to game the system, too, as computers have improved and algorithms have gotten more sophisticated. And now there’s AI.
Does this mean Delta will know when you’re flying to a funeral on short notice and charge you full fare, because it knows you’ll pay it? Probably not, according to Laurie Garrow, a professor of aviation at George Tech.
“The characteristics that they’re looking at to do this discounting are characteristics about your trip — how far I’m booking in advance, what market, how many people are traveling together, historic purchase patterns in the aggregate — that’s what’s being fed to the AI,” Garrow said. “It’s not things that are being hypothesized, like, ‘Am I going to a funeral?’”
In other words, what Delta knows about is based on the data you’ve given it, especially when shopping for flights.
It turns out that what’s happening with Amazon prices is a little less nefarious. The company uses dynamic pricing with the help of algorithms — it constantly changes prices based on supply and demand — but Amazon says it doesn’t do personalized or surveillance pricing. So that super low price on boxer briefs wasn’t just for me. Amazon reportedly changes its prices up to 2.5 million times a day, or about once every 10 minutes per product, but those fluctuations show up for all customers.
“Algorithmic pricing exploits a vast asymmetry of information,” said Elise Phillips, policy counsel at Public Knowledge. “Given that there is often little transparency in how these algorithms function, it’s fundamentally unfair to consumers.”
What you can do to avoid the algorithm’s gaze
It may feel impossible to escape the reach of dynamic pricing both online and off. Again, it’s not just airlines and Amazon doing it.
Ticketmaster, which is infamous for dynamic pricing, and its parent company Live Nation are being sued by the Federal Trade Commission (FTC) for illegally operating a monopoly. (So is Amazon.) Kroger is being scrutinized not only for its personalized discounts but also its new electronic shelf labels, which change the prices of groceries for arbitrary reasons. Uber has normalized surge pricing, and is now finding new ways to overcharge customers.
If you do find yourself shopping for a flight, a pair of underwear, or even a concert ticket, and you want to be treated like a stranger, make yourself a stranger online.
You can avoid spending money with companies who use dynamic prices, if there’s a good alternative. (For air travel, unfortunately, that’s nearly impossible because most airlines use a form of dynamic pricing.) Nevertheless, if you do find yourself shopping for a flight, a pair of underwear, or even a concert ticket, and you want to be treated like a stranger, make yourself a stranger online.
You can do this several different ways. Use a VPN when you shop online. And then try it using incognito mode in Chrome or a private browser window in Safari to keep data out of the algorithms’ hands. You could even use a privacy-focused browser, like Duck Duck Go, to avoid even more trackers. Heck, try all three approaches and see if the prices change. It is theoretically possible that surrendering your personal data could lead to lower prices. That doesn’t make it right, though.
“Just because the technology exists, we don’t just have to let companies do what’s most profitable for them,” Brian Callaci, chief economist at the Open Markets Institute, told me. “You know, the whole point of capitalism should be working for us and not the other way around.”
A version of this story was also published in the User Friendly newsletter. Sign up here so you don’t miss the next one!
