The AI Bill of Goods Hitting Schools Right Now

Somewhere tonight there's a school board meeting happening in a room that smells like burnt coffee and carpet cleaner, and someone is standing in front of a projector showing a slide deck about how artificial intelligence is going to transform education, and the administrators in the folding chairs are nodding with the specific energy of people who are terrified of being the district that didn't try. I've been in nonprofit IT long enough to recognize that nod. It's not enthusiasm. It's survival instinct dressed up as innovation.

The Wasabi survey, which the vendors will absolutely not be citing in those presentations, tells a story worth sitting with: 46% of education organizations plan to increase AI spending, 98% expect their budgets to hold or grow, and only 37% of current AI projects are showing a positive return on investment. If you run that math in any other industry, the person who approved those projects is updating their resume. In education, 37% gets you a follow-up grant application and a mention in the superintendent's annual report about forward-thinking initiatives. The metric isn't whether it works. The metric is whether you did something, whether you can show the board you're keeping pace with the district down the road that got a write-up in EdWeek for their AI pilot. The actual outcome is secondary to the appearance of motion, and that's not a bug in the system, that's the system.

Then there's the part nobody puts in the slide deck: fee-related charges (data egress, API operations, access fees) now account for 54% of cloud storage spending in education, up from 50% the year before. Because the hyperscalers don't make their money on the storage you use, they make it on the movement you can't avoid. Every time you want your own data out of their parking lot, you pay the toll. Sell the shovel cheap, charge for every swing. I've watched this trap close on nonprofits and I'll tell you exactly how it works: the initial numbers look clean, the grant has a use-by date, the board wants to see something in the annual report, so you buy in. And then the real costs start arriving, the egress fees, the API calls, the storage tier the sales rep definitely mentioned but not like that, the migration costs when you eventually want to leave. By the time you understand the pricing model you're three years deep and the switching costs are higher than the overages. That's not an accident. That's architecture. Designed to keep you in place (and I'm not calling it malicious, but I'm not calling it benign either).

The thing that actually keeps me up is not the budget line itself, it's what that budget line is displacing. Every dollar allocated to AI infrastructure is a dollar that isn't going to a teacher's salary, a building repair, a library that stays open one more afternoon a week, a bus route that doesn't get cut. The unglamorous stuff. The stuff that doesn't come with a keynote or a press release, that's just maintenance, just care, just the grinding daily work of educating actual human beings in actual buildings that have actual roofs that leak. Somehow that's not enough to justify a line item, even though it's the whole point. Even though it's what the kids actually need. The shiny thing gets the grant and the old thing gets deferred until it fails catastrophically, and then we wonder why the system is struggling.

I don't have a clean policy fix for this and I'm not going to pretend I do. The incentives are misaligned at every level: vendors get paid whether the AI delivers or not, administrators get credit for trying, teachers get handed another tool nobody asked them about and told to integrate it into their already-overloaded days, and students get whatever's left over after the budget has been carved up by things that sound impressive in board meetings. But maybe the next time someone stands up with a PowerPoint and a promise, somebody in the room asks the questions that don't have slides. How much does it cost to get the data back out? How many teachers does this budget cover? What happens in year three when the grant money is gone and the licensing fees aren't? That's not a revolution. It's just arithmetic. But nobody's doing it yet, because doing it would require caring more about the kids and the teachers than about being the district that was first.