The playground from a school director of innovation’s childhood featured a metal slide that stood roughly 12 feet high, positioned straight up and straight down. There were no curves, no protective plastic tubes—just a steep, unyielding sheet of metal accompanied by a ladder leading straight to the top. On occasion, the director’s father would bring along glass cleaner, spraying down the slide to make it slicker. The children would fly down at breakneck speeds, launching themselves off the end into open space.
It was a beloved childhood playground, one that has increasingly resurfaced in memory during board meetings and curriculum planning sessions. The professional context is vastly different today: as a school director of innovation serving a population of boys in fourth through ninth grades, the daily challenge is figuring out when, how, and if to introduce artificial intelligence to the classroom.
That same playground also featured a merry-go-round of the old-school variety, powered entirely by the raw energy of children. A few kids would run alongside it as fast as they possibly could while everyone else clung on for dear life. Eventually, the runners would leap aboard, testing their grip to see how long they could last before centrifugal force sent them flying off into the grass.
It would be easy to twist these recollections into a tired morality tale about how children used to be tougher, but that misses the entire point. Children genuinely got hurt on playgrounds of that era. Equipment featured hidden pinch points and dangerous entrapment hazards, and kids regularly fell from tall structures onto unforgiving concrete surfaces. By the 1970s, mounting federal injury data prompted the introduction of new safety guidelines from the Consumer Product Safety Commission, transforming the American playground landscape forever.
As playgrounds became systematically safer, however, researchers began asking a deeper, more probing question: Had well-intentioned adults sometimes inadvertently confused the elimination of physical hazards with the complete elimination of risk?
As it turns out, children may actually derive profound developmental benefits from play experiences that involve height, speed, uncertainty, and even a touch of fear. A hazard is broadly defined as something a child cannot reasonably be expected to recognize or know how to handle. Managing risk, by contrast, asks them to learn and grow by making their own judgment calls.
For the past several years, a central part of the job at a Massachusetts school serving boys in grades four through nine has been navigating precisely where artificial intelligence belongs in the educational ecosystem. The school has developed internal guidelines, experimented with AI in controlled, supervised environments, and spent countless hours in deep discussions with teachers regarding when these emerging tools genuinely support learning and when they might inadvertently get in the way.
Yet, what once felt like a manageable set of pedagogical questions has steadily grown more complex. Lately, the decisions seem to be getting harder, not easier.
Powerful Signals From Large Districts
The national conversation shifted dramatically at the start of the school year when New York City Public Schools enacted a ban on the use of student-facing generative AI through the eighth grade. The Los Angeles Unified School District went even further, implementing a strict policy prohibiting all student use of generative AI on district-owned devices for the entire academic year.
Those high-profile policy decisions fundamentally altered the tenor of the conversation at schools across the country. Teachers who had been thoughtfully trying to figure out how middle school students should appropriately wield artificial intelligence were suddenly forced to confront a much more fundamental question: Should middle schoolers be using it at all? When two of the largest, most influential school systems in the United States decide to pull back and hit the brakes, educators everywhere take notice, particularly when those districts are wrestling with children of the exact same age group.
At the same time, parents are absorbing other reasons to exercise deep caution. Families are increasingly exposed to media reports about increasingly capable artificial intelligence systems behaving in ways their human developers never intended. This summer, for instance, an incident involving experimental AI agents demonstrated that these systems could circumvent built-in safeguards, find unauthorized ways to communicate when they were supposed to be completely isolated, and gain unauthorized access to external systems.
Admittedly, these are complex enterprise-level systems rather than the tame educational chatbots a middle school student might encounter in a classroom setting. It is an important and valid technical distinction. Yet, any educator or administrator can readily understand why that distinction only goes so far for a parent sitting at a kitchen table, trying to decide what kind of technology belongs directly in front of their child.
Meanwhile, just a few states away in Massachusetts, the regulatory and cultural signal looks remarkably different.
A Different Message From Massachusetts
Rather than implementing a sweeping temporary ban on artificial intelligence for all students or drawing a hard statewide grade-level line, Massachusetts has actively encouraged individual schools and local districts to develop comprehensive AI literacy skills and design thoughtful, localized policies around student use. Furthermore, the state is actively supporting an artificial intelligence curriculum pilot tailored specifically for grades 8 through 12.
That striking contrast in philosophical approaches has lingered in the minds of educators trying to chart a sensible course. While an eighth grader sitting in a New York City classroom is navigating the final grade covered by a strict one-year moratorium, an eighth grader in Massachusetts is stepping into a classroom where a state-supported initiative to explicitly teach students about artificial intelligence is just beginning.
Certainly, there are vital, substantive differences between teaching foundational AI literacy and simply handing a student loose access to an open-ended chatbot. Even so, these represent profoundly different institutional responses to the exact same unsettled, overarching question: What do children actually need from us right now?
Versions of this difficult, highly nuanced debate echo constantly within local school halls.
Some veteran teachers wonder aloud whether students are encountering artificial intelligence before they have had adequate time to develop foundational skills without its crutch. Some parents question how much confidence anyone can reasonably place in tools that are evolving at such a breakneck pace. At the same time, educators are deeply aware that these exact same students are inevitably headed toward a professional and personal world where entirely avoiding artificial intelligence will become increasingly impossible.
Everyone involved in the discussion is ultimately trying to protect the exact same children. The underlying anxiety is uniform; the disagreement lies entirely in which aspects of an uncertain future pose the greatest threat.
There are undeniably legitimate reasons to worry about moving too quickly. What happens to the fundamental art of writing when producing competent, grammatically correct prose becomes utterly effortless? What happens to deep, critical thinking when immediate answers arrive before students have even learned how to formulate a proper, thoughtful question? What happens when a young person becomes conditioned to turn instantly to an external tool whenever an academic assignment becomes difficult?
Conversely, waiting on the sidelines carries its own distinct set of risks. If artificial intelligence becomes foundational to how human beings work, create, communicate, and solve problems, what happens when students leave the K-12 education system completely unprepared and lacking any fundamental understanding of the technology? At some point, young people require structured opportunities to develop sound judgment while experienced teachers and parents are still readily available to guide them.
Neither of those concerns is imaginary or hyperbolic. Both represent very real stakes in a rapidly shifting educational landscape.
Perhaps that is why the memory of that old playground keeps surging back to the surface.
Addressing Safety Concerns
As noted previously, there were genuine dangers on playgrounds like the one from decades past—aspects that children certainly should not have been expected to navigate on their own. Exposed pinch points, dangerous entrapment hazards, and hard concrete sitting directly beneath towering climbing structures. Adults quite rightly stepped in to address those specific physical safety concerns. But those adults did not look at the risks and decide that children should never be allowed to climb high again.
There is a profound distinction embedded in that history that modern schools must consider when evaluating technology.
Perhaps properly preparing children for an artificial intelligence-saturated world does not actually require saturating their delicate childhoods with artificial intelligence.
A seventh-grade student who does not interact with a generative chatbot this academic year will almost certainly get another opportunity down the road. Artificial intelligence is not going anywhere.
However, there are certain formative experiences that students should not postpone or bypass. Wrestling with a challenging question before an instantaneous answer appears on a screen. Discovering their own unique voice on a blank page before a machine offers to optimize and rewrite it. Working through academic and intellectual confusion directly with another human being. Being bored long enough for natural curiosity to take over. Finding out what their own unassisted mind can accomplish before something else offers to do a portion of the heavy lifting for them.
None of those essential developmental experiences sound particularly futuristic. Yet they are fundamental to how human students learn to think independently. That is precisely where their enduring value lies.
Balancing Safety With Managing Risk
At the same time, students undeniably need to learn how to use artificial intelligence well and ethically. Keeping them entirely away from a consequential technology carries its own heavy risks. They will ultimately need to understand what these complex systems can actually do, where they inevitably fail, how they subtly shape the information they encounter, and when relying upon them constitutes a poor choice.
The central question teachers and administrators must answer is clear: What does a school genuinely owe a child when the world operating outside the school walls is changing far faster than society can decide whether those changes are ultimately good for them?
Eventually, educators have to let students onto the playground. The exceptionally difficult decision is determining the exact right moment.
Schools are currently being asked to make these high-stakes decisions while the very technology being evaluated is still actively transforming beneath their feet. The artificial intelligence that students encounter one year from now may bear little resemblance to the AI currently being debated in faculty rooms and school board meetings. Educational research takes years to catch up and analyze impact; childhood, by contrast, does not wait.
Students are walking back into classrooms every single morning. School leaders have to make concrete choices for them right now, fully aware that some of those decisions may look markedly different with the clear benefit of hindsight.
Decades ago, a young kid stood at the very top of a towering metal slide, completely unconcerned with federal safety standards or whether launching off the end was developmentally appropriate. That was somebody else’s job to worry about.
Now, that responsibility belongs to the next generation of educators. And the view from the top looks remarkably different.