AI companies are running a play we’ve seen before. They’re telling schools that AI is the hot new skill, the place where all the jobs are, and that kids who don’t learn it now will fall behind. To help, they’re offering curriculum and tools, often for free. That’s the pitch, and according to The Verge AI, which spoke with New York Times education technology reporter Natasha Singer on a recent Vergecast, it’s nearly identical to how the tech industry embedded itself in classrooms over the past 15 years.
Singer’s new book, Coding Kids, traces how the last cycle worked. Companies promised computer science skills were the ticket to high-paying jobs. Schools helped build a pipeline of future workers, and those workers got trained on specific products and became lifelong customers. By the time many of those students graduated, the job market had shifted and entry-level coding roles were shrinking. What stands out here is her warning that we’re about to repeat it. “I worry we have collective amnesia,” Singer told The Verge AI.
The playbook, then and now
The 2010s version had a few reliable moves:
- Own the curriculum. Apple and Microsoft each built their own AP Computer Science Principles courses, featuring their own tools like Swift and Minecraft. Singer’s sharp comparison: parents would never accept “Pfizer AP Biology” or “ExxonMobil Environmental Science,” yet tech-branded coursework slid in without much notice.
- Own the hardware. Google put low-cost Chromebooks in classrooms across the country, a rollout that accelerated during pandemic shutdowns. Add Google Classroom for assignments and grades, and by the time generative AI arrived, Google was already the default platform in many districts.
- Fund the movement. When companies stayed at arm’s length, they backed nonprofits like Code.org, whose celebrity-packed “learn to code” video featured Bill Gates and Mark Zuckerberg. The “Hour of Code” campaign followed.
The AI industry is now pulling from the same three-part script, with the same urgency.
What’s different this time
Here’s the twist Singer is optimistic about: the pushback showed up early. In the coding era, critics were few and far between. Not now.
- New York City, the nation’s largest school system, just banned AI use in elementary and middle school classrooms for the coming year.
- Los Angeles went further a day later, extending restrictions to high schoolers.
- Some districts are pulling out Chromebooks, and global research keeps finding that piling tech into classrooms often doesn’t move actual learning.
There’s a grassroots movement of parents and teachers fighting the screen-ification of school. Singer’s read: “I think schools are beginning to figure it out.”
Why it matters now
The core question Singer raises is whose question schools are answering. “How can we get AI into schools faster?” is a product question. “What are the most important things we want kids to learn?” is an education question. When the industry writes the curriculum, the product question wins by default. That’s the trap, and it’s being set again right now, while safeguards are still cheap to install.
What to do about it
For educators, district leaders, and the AI companies courting them, a few practical takeaways:
- Treat vendor curriculum like a conflict of interest, not a gift. Free lessons that teach a specific product create tomorrow’s locked-in customers. Ask who benefits.
- Separate the tool from the thinking. Kids can learn to question how companies steer users with technology, not just how to click the buttons.
- Demand evidence. If tech in the classroom doesn’t improve learning outcomes, that’s the metric that should decide adoption, not job-market fear.
- For AI vendors: the districts saying no this year are a signal. The companies that build genuinely independent, evidence-backed tools will age better than the ones optimizing for classroom market share.
Over the next few years, expect this fight to sharpen. Districts that moved fast in the coding era are now unwinding those choices, and they’ll be slower to hand AI companies the same keys. The lesson from the last cycle is simple: the safeguards are easier to build now than to retrofit a decade from now. You can find the full conversation with Natasha Singer at The Verge AI.