Airline Tickets and AI: A Cautionary Tale about Automated Pricing Practices

The business news pages have run several stories in the last month about a group of US senators demanding answers from Delta Airlines about its plans to use AI to set ticket prices.  Regardless of the airline's actual plans, the rash of coverage gave plenty of airtime to consumer advocates and privacy champions, put the company into a defensive posture publicly, and may spur ongoing inquiry into ticket pricing practices.  The issue stemmed from prior reported statements by a Delta official that AI could help the airline charge “the amount people are willing to pay” for a particular ticket.  Reports of that statement set off a flurry of questions from senators, prompted proposed legislation in the House, and invited critical commentary from the Secretary of Transportation.  

WHY IT MATTERS

“Dynamic pricing” is not new; neither is the use of revenue management tools and software.  Retailers are known to use pricing and revenue tools to automate price-setting based on factors such as supply and demand or overhead costs.  In recent years, however, large online retailers (such as Amazon) have been the subject of speculation about how they set prices for individual consumers, and whether personal data is used to drive prices based on an individual's perceived ability or willingness to pay more for a particular purchase.  In the case of airline ticketing, that could mean (to use one reported example) raising the price of a flight if the consumer has just viewed an online family obituary notice.  Delta has denied that it uses personal data to set prices, but the industry is facing ongoing questions about automated pricing practices.  

The US has no national laws regarding retail use of consumer personal data, although approximately half the states have consumer protection laws that limit personal data use to various degrees.  The undisclosed use of personal information to raise prices would likely run afoul of at least some of those laws.  The same is true of emerging state AI laws, some of which will require disclosure.  

The lesson to draw from all this?  If you use automated pricing, particularly if that pricing is set via an AI, consider disclosing the practice.  Telling customers what data are used in those automated decisions, and what data are not used in them, could help quell confusion and potential backlash.  Although the US appears to be adopting a light-touch regulatory regime regarding AI use, PR and other considerations could still dictate that disclosure is the best practice.