The Funding Pulls One Way. The Generation Pulls the Other.

On April 13, the U.S. Department of Education quietly finalized a rule that, starting May 13, will weight federal K-12 grant dollars toward applications that put artificial intelligence deeper into classrooms.¹ The priority is broad. Proposals that integrate AI literacy into teaching practice, embed AI lessons into teacher preparation, fund professional development for educators, support adaptive tutoring tools, or use AI to lighten administrative load will all stand a better chance of being funded.¹²

Three weeks earlier, on April 9, the Walton Family Foundation, Gallup, and GSV Ventures released the latest "Voices of Gen Z" survey.³ It is the most reliable read we have on what young Americans actually feel about AI right now. The fieldwork ran from late February through early March. The sample was 1,572 people between 14 and 29.³

The numbers are striking. Anger about AI rose nine points in a single year, to 31 percent. Excitement fell fourteen points, to 22 percent. Hopefulness fell another nine points, to 18 percent.⁴ Only 46 percent now believe AI will help them learn faster, down from 53 percent the year before.⁴ Eight in ten Gen Zers said it is likely that using AI will make it harder, not easier, for them to learn over time.³

These are not the views of refusers. About 51 percent of the same young people use generative AI at least weekly.⁴ They are not opting out. They are using the tools and watching themselves use them, and they do not entirely like what they see.

A Federal Priority Built on a Premise the Students Question

The Department of Education's new rule is, in form, neutral. It does not prescribe a particular tool or platform. It simply tilts the next several years of grant competitions toward AI-forward proposals.¹ Districts that want federal money will read the priority closely. Many will reshape their applications to fit it. Some will adopt tools they were not otherwise planning to adopt.

That, by itself, is not a problem. The problem is that the underlying premise of the priority, that more AI in classrooms produces better student outcomes, is exactly the premise the surveyed students are now contesting in measurable, year-over-year shifts. They are not saying AI is useless. They are saying it is doing something to their thinking that they cannot quite name and do not entirely trust.

Federal policy and student sentiment are running in opposite directions. The two trains have not yet collided, but the tracks are converging.

The Quiet Variable Is Visibility

It is tempting to read the survey as a verdict on AI itself. It is more honest to read it as a verdict on a particular shape of AI. The kind that arrives as a black box, returns a finished answer, and leaves no record of what the student was thinking before they typed the prompt.

That shape of AI is the one that earns the suspicion. The student who pastes an essay prompt into a chatbot, copies the result, and turns it in does not feel proud. They feel a little smaller. They feel that something was outsourced that should not have been. The Walton survey is, in part, a measure of how many students have had that feeling enough times to start naming it.

A different shape of AI is possible. One that sits next to the student rather than in front of them. One that watches the writing happen and notices the pause before the third paragraph, the sentence that was rewritten four times, the moment a vague idea sharpened into a clear one. One that gives the teacher a record of how the thinking moved, not just what the final answer was.

This is the work we do at Koan. Aidan, our AI tutor, is built to make learning visible. Every revision, every hesitation, every breakthrough is captured and rendered legible to the student, the teacher, and the parent. The thinking is the product. The answer is just where the thinking happens to land.

A federal grant priority does not have to choose between more AI and less AI. It can quietly favor the kind of AI that returns the work of thinking to the student rather than performing it for them. The current priority does not yet make that distinction. The students surveyed in February seem to be asking, in their own way, that someone start to.

The Window Is Short

The new rule takes effect on May 13. Grant applications written under it will shape what reaches classrooms over the next several years. The students who will live inside those classrooms have already given their early review. They are skeptical, and they are paying attention.

The honest task for districts, vendors, and policymakers is to design AI that earns the trust the survey is no longer giving freely. That cannot happen with louder marketing. It can only happen by making the actual work of learning visible enough that a student can see, with their own eyes, that they are getting smarter rather than less so.

If the next round of grant applications had to prove not that AI was used, but that learning was visible, what would the winning proposals look like?