KENNESAW, Ga. | Nov 12, 2025

While most of us know that we need to be diligent about validating what we read on the Internet, it is easy to slip up and accept something as fact without checking it. Even experts can be fooled.

In this blog, I’ll discuss a recent example of how I got fooled by inaccurate reporting. I incorrectly took airline pricing practices described in widely distributed articles at face value. I was disappointed to realize that my own post referring to the articles amplified the false information and became part of the problem!

Yup, I Fell For Fake News

This past summer, I saw some articles about how Delta Airlines was supposedly engaging in “surveillance pricing”. That means fares being dynamically set based on predictions about what an individual is willing to pay at a given moment (see here). If true, this could lead to all sorts of nasty situations such as gouging grieving relatives who must go to a funeral or taking advantage of grandparents planning a once-in-a-lifetime trip for their grandkids.

In reaction to the articles, I posted on LinkedIn (see here) and tore into the practice. I even ran a poll (see image below) where dozens of people chimed in. While the post got a moderate number of views compared to what I typically see, it still had ~3,000 impressions. That means that my amplification of false information spread well beyond myself.

Even though I know how traditional airline pricing systems work and even though what I read was in opposition to some of the practices I knew were long standing, I allowed my annoyance with the claims to cause me to run with the narrative.

What I Learned Was True

After taking a lot of heat, Delta clarified that they have no intention of using surveillance pricing (see here and here). Rather, they are going to use AI to further automate their classic pricing and revenue management approaches. For example, allowing AI to adjust available fare classes in near real time. After more research, I realized the truth was a classic analytics or AI use case: automating and scaling processes that traditionally relied heavily on human involvement. I posted a follow up acknowledging my error (see here). It was also seen by ~3,000 people, but many of those 3,000 were certainly not the same ones who saw my initial post. Thus, there are plenty of people I misled that didn’t see my correction.

With that said, Delta isn’t entirely blameless here. As often happens, a non-technical executive talked excitedly about a new technological capability without fully accounting for the nuances. The way some of the original statements were worded in Delta’s presentation left room to interpret them as something they were not, and some reporters interpreted them in the worst possible way while apparently not digging for verification before publishing. Next, a lot of people, including data and AI experts like me, ran with the incorrect narrative.

For context, realize that airlines have a range of fare classes with prices attached. The airlines don’t often change those prices but instead shift which fare classes are available at any point in time. Even if they wanted to price individually, consider the high percentage of price searches that come from 3rd party travel sites. On those sites, it is not possible to personalize prices since airlines wouldn’t know who you are. In addition, a standard practice is for travel sites to pull from a central flight pricing database. That database can be updated frequently by an airline for any given flight as conditions change, but anyone who asks for a price at a point in time will see the same price.

Even if airlines could personalize pricing on 3rd party platforms, it wouldn’t fly with consumers. I can accept seeing a different price tonight than this morning but not seeing different prices on different sites at the same moment. Thus, aside from technical hurdles, consumers would revolt if surveillance pricing was implemented. Beyond those barriers, there are also contractual agreements on commissions and pricing consistency between airlines and 3rd party sites. Those agreements would also pose legal hurdles to surveillance pricing implementation. Bottom line: surveillance pricing is not happening any time soon!

The Lesson And Actions To Take

The point of this story is to remind readers that it is very easy to misspeak, to misunderstand what someone else says, or to believe what we read. With highly charged and sensitive topics like AI, surveillance pricing, privacy, and ethics, it is easy to jump to conclusions and assume the worst … especially when all those topics collide in a single situation.

I am embarrassed that I fell for the Delta story and I should have dug deeper before making my post. I knew enough to recognize that the practices claimed in the story had major barriers to implementation, deviated widely from past practices, and simply didn’t pass my sniff test. At the same time, my sensitivity about privacy and ethical violations led me to override my gut.

Each of us should be careful to diligently assess information about the use of AI and data (and the same should apply well beyond just those topics). If we can validate there is a real issue, then absolutely we should call it out. However, simply believing what you read in an article or two isn’t enough … especially given that many of today’s articles rely on AI to provide some of their “facts” and many authors are using their articles to further wider agendas. If our guard is down, even experts can fall for fake AI news. Don’t be one of them like I unfortunately was in this case!