Picture two learners and one technology. In the first classroom, the adults teaching share a clear, written agreement about how AI may be used and how work will be judged. Students there learn to use the tool in the open: they draft with it, then defend their choices, and they know where the line sits. In the second classroom, there is silence from the top, so every teacher improvises a policy of one. One bans AI outright and hunts for it with a detector. The teacher next door quietly allows it and never says so. A third has students paste raw AI output and calls it research. Same tool, three different educations, and the learner did not choose any of them.
This is what an absence of design looks like at the level of a system. Nobody decided that a student’s experience of AI should depend on which door they walked through. It happened because the decision was left unmade, and unmade decisions still have outcomes.
A patchwork is a design, too
When shared guidance is missing, something still fills the gap. Usually it is procurement. Whatever tool got purchased, with whatever defaults it shipped, quietly becomes the working policy. Buy a plagiarism detector and you have decided, without a meeting, that AI is a policing problem. Buy a chatbot that writes lesson plans and you have decided the opposite, again without a meeting. The purchase order is the pedagogy, and almost nobody experiences it as a choice.
When guidance is absent, the purchasing decision quietly becomes the pedagogy.
The pattern is old. Cuban’s study of computers in classrooms found that machines arrived with big promises and then got bent to fit whatever the local setting already did, which was often very little (Cuban, 2001). Reich tells the same story at the scale of national platforms. Distribution scaled fast, and the learning it was supposed to carry did not follow on its own (Reich, 2020). The tool does not decide what happens. The design around it does, and when the design is left blank, the least-resourced setting fills the blank with whatever it can manage.
Who pays for the gap
The cost of a patchwork is not shared evenly. A student with a parent who understands AI, or a well-funded school with time for professional learning, absorbs the confusion and moves on. A student without those advantages absorbs the confusion and stops there. The same uneven guidance that is a minor inconvenience in one place is a real barrier in another. That is what makes this an equity story and not just an administrative one. Clarity is not a nicety here. It is the thing that decides who gets a fair shot at using a powerful tool well.
The design response
The fix is not a better tool. It is clarity that someone actually owns. Shared, plain expectations about where AI helps and where it gets in the way. Assessment redesigned so that thinking stays visible, which makes the question of cheating mostly moot. And real support for the people doing the teaching, so the guidance survives contact with a Tuesday. Uneven design produces uneven learners and then calls the difference a difference in ability. Naming that is the first honest step. Fixing it is a design job, and it belongs to whoever drew the map.