Over the past few years, the term “synthetic data” has come up a lot, as planners think about how to source inputs for AI engines. But in reality, there’s no substitute for the real thing.

The problem is: where to get all of the data that’s needed. Many remember the spate of lawsuits against Anthropic, for pirating book data, some of which are still going on. That’s an example of where not to get information, in order to avoid triggering massive legal conflicts and miring down your company in the process.

Google has a different take: new reports show the AI mega-firm behind Gemini has bought enormous amounts of data from Spirit Airlines, as part of a bankruptcy sale. Google will pay around $10 million, but that turns out to be bargain basement prices, especially compared to the costs of litigation from pilfered data.

What is Google getting? Taylor Delandro at News Nation reports that the data consists of “roughly 10 million emails and 500 million Microsoft Teams chats, as well as other documents, spreadsheets, calendars and other internal company data.”

De-Identifying the Assets

To allay fears of privacy violations, Google is apparently agreeing to “scrub” records for personal information. In addition, certain components of the data set, like nearly 100 million passenger profiles and some 50 million records related to the former airline’s Free Spirit loyalty program will not be included in the exchange.

That’s all in aid of making sure that the use of this data to power AI doesn’t encroach on the rights of those who flew Spirit in the past.

Other coverage of the plan shows what’s happening to the rest of Spirit’s assets. The planes, for example, will cool their heels in the desert of Arizona , where the weather is more hospitable to their corrosion-prone bodies.

“Most of the time, they go into storage where they can be maintained,” explains Bob Allen of plane leasing firm Nomadic Aviation Group, in a piece by Ashley Soriano showing some of the logistics around Spirit’s shutdown. “The air is dry and hot out here, so there’s less chance of corrosion on the aircraft while they’re sitting.”

As for the employees, reporting shows they will be able to use a platform called Kitty Hawk to network and try to find other jobs in their field.

“I’m trying to connect all of aviation. Pilots, products, services, companies,” Allen said, in the above article. “It’s in its infancy, but it’s very helpful to any pilots, any employees that are looking for work right now can go on there, leverage AI to develop their networks. The difficult deal is the guys that are 50 plus years old, guys like me that have lost their job and now have to go look for something else, that something else is starting completely over because it’s a seniority-based system out there.”

That’s it in a nutshell for a lot of people, whether they were employed at Spirit or anywhere else that either let them go, or imploded in a fierce maelstrom of late-stage capitalism. In a tight job market, there’s not a lot of lateral mobility.

It turns out that a runner-up bidder in the process was Mercor, itself a networking and recruiting firm, that would have used that real data to those ends. It shows that these sorts of data sets are valuable on the market, useful to various companies that are racing for the newest solutions, like children under a pinata.

Here’s a demonstrative blurb from Mercor’s own web site:

“Frontier AI labs can't build enterprise-grade models from public data alone. The next leap in AI capability requires operational data: how teams communicate, close deals, ship software, and run finance. That data lives inside your company.”

But that hot auction also gives away the notion that firms are chasing “real data.”

I wrote recently about a similar indicator, where a company auctioning off Nvidia GPU capacity got multiple bidders chucking the end price upward. That’s when you know you have something.

In any case, what we know is that, in using the data of the defunct airline, Google will not be subject to various types of liability. Because they did it the honest way. They bought the data fair and square.