Last month, Delta President Glen Hauenstein announced to investors that Delta would expand its pilot program that uses artificial intelligence (AI) to maximize what individual consumers pay.
This has brought new attention to the potential for sellers to adopt bespoke prices, tailored to each customer based on their personal data.
Delta has subsequently insisted that it is not using AI to determine individual airfares, but merely to enhance the ways it has traditionally used dynamic pricing to adjust airfares to meet supply and demand generally. Though it is difficult to see how those traditional market pricing metrics could be so improved using AI as to warrant the accolades Hauenstein shared.
Still, concerns remain that targeted AI-powered pricing models will enrich sellers at consumers’ expense. To fix the enormous information disparity between sellers and consumers that enables this intrusive practice, we must fix the lack of consumer privacy protections in U.S. law. And we must also empower consumers to protect themselves.
The problem with bespoke pricing
An unbounded collection of consumers’ personal data, along with advances in digital technology, is enabling sellers to use powerful algorithms to invade a consumer’s privacy and capitalize on the results. Sellers can create an intimate profile of each consumer, and use it to assess that consumer’s unique willingness, and ability, to pay more than the price that the traditional supply and demand curves would yield.
For more than a century, sellers have posted a list price for their wares, available uniformly to all consumers, who are individually anonymous to the seller when the price is posted. Sellers have used market research to determine demand and adjust the price accordingly. Consumers who would have been willing to pay more than the list price get the benefit of what economists refer to as the “consumer surplus.”
With bespoke pricing, the seller turns the tables on consumers. Corporations will be able to obscure the prices they offer; meanwhile the individual consumer will be an open book. Over time, sellers can reduce the consumer surplus to zero.
Read More: What Happens When AI Replaces Workers?
Theoretically, bespoke pricing could result in prices being reduced for consumers who want the product or service but not at the list price. Budget-conscious consumers would benefit, subsidized by consumers who are willing and able to pay more. Indeed, to some extent, this already occurs in many parts of our economy. For instance, consider how some consumers are willing to pay more to have their online shopping order expedited, while more budget-minded customers might wait for a seasonal sale with standard shipping.
But sellers do not offer lower prices to some consumers altruistically. Their goal is to increase sales and reduce excess inventory, increasing overall profit. They have traditionally accomplished this by holding sales. But with whatever is the current price transparent to all, consumers can time their purchases, anticipate sales, and shop around. That approach allows consumers who would be willing to pay more, but who wait for or happen upon the lower price, to grab the consumer surplus. Bespoke pricing takes away this possibility, enabling the seller to seize that consumer surplus for itself.
For decades, airlines have taken advantage of what they know about their market to create different categories of seats at different prices, based on such factors as time of year, time of purchase, and class. They anticipate that demand will increase during holiday seasons or around big events. They infer that most consumers purchasing well in advance are likely to be individuals and families traveling for vacation or other personal reasons, who generally try to be budget-conscious. They infer that most consumers purchasing close to the time of travel are likely to be traveling for business reasons, often on a tax-deductible corporate expense account, and therefore willing to pay more. And consumers who want to fly first class are in a category by themselves where, relatively speaking, money is no object. Airlines also pioneered the loyalty program, offering rewards to frequent flyers.
But those differentiated prices are publicly advertised and available for all to see.
Most flights are at or near capacity. Realistically, no airline is going to be able to confidently assess how many seats will remain unsold early enough for offering individualized discounts to consumers to be profitable. The usefulness of bespoke pricing to an airline’s bottom line will instead be to exact higher prices from consumers whose profiles show a willingness to pay more.
An AI-powered ‘invisible hand’
Of course, individualized pricing isn’t always unfair. In medieval bazaars where a seller could size up a buyer and a buyer could likewise size up a seller, completed sales were the product of negotiation. Some major purchases, like homes and cars, still often occur this way.
But generally, individualized negotiations proved inefficient for a modern industrialized economy. Beginning in the 1880s, posted and advertised list prices—set before sellers knew who would be buying—became the norm. The price tag ushered in anonymity for shoppers.
When online commerce arrived, consumers had reason to hope that it would improve their purchasing power by making it easier for them to shop around and find products and services they want at the best price—and indeed it did. But the advent of bespoke pricing is threatening to betray that hope. Online commerce also engendered a new market for consumer data, which has been collected, organized, distributed, sold, and cross-referenced on a massive scale, enabling sellers to fundamentally reverse the anonymity equation. When powered by AI, Adam Smith’s “invisible hand” may soon be picking the consumer’s pocket.
Sellers can now access personal data including the consumer’s previous purchases or searches; income, assets, debts, and financial condition and history; personal and family life; employment, work life, and career history; political, social, and other activities the consumer and the consumer’s family members and friends have engaged in; web-browsing and social media history; and broadly, anything about the consumer recorded or tracked and fed into the big data maw.
Read More: A Potential Path to Safer AI Development
What can consumers do to not surrender their power and personal data?
One possible recourse is to convince their government to rein in the practice of bespoke pricing. Last year, the Federal Trade Commission investigated it. Now, legislation is under various stages of consideration in Congress and in state legislatures.
Enacting stronger data privacy laws is an important step.
Without regulatory limits, another possible recourse for consumers is to take the matter into their own hands, by obscuring their identity and limiting profiling by sellers. For example, they might shop through an online virtual private network (VPN).
Bespoke pricing is turning the tables on consumers, and returning some power to them will be challenging. Consumers and their public interest advocates are up against the vastly superior financial, technological, data, and lobbying resources of industry. But one way or another, we need to confront bespoke pricing, so that the online marketplace will still work for consumers.
The post The Issue With AI-Powered Pricing appeared first on TIME.
