The Trouble With Tech Abstinence

Phone bans make sense. But AI may solve some of education’s hardest problems while creating entirely new ones.

A major recent study from the National Bureau of Economic Research (NBER) examined one of the most popular school reforms of the moment: banning student cell phones. Using nationwide data from thousands of schools, researchers found that strict bans sharply reduced in-school phone use. But the academic results were far less dramatic than proponents likely anticipated or expected. On average, test scores barely budged.

That finding doesn’t mean phone bans are misguided. It means we may be misunderstanding the problem they actually solve.

Phone bans remove an obstacle to student attention, and that’s intrinsically valuable. Schools shouldn’t need dramatic gains in test scores to justify removing a constant, engineered distraction from the classroom. A smartphone is not a neutral tool; it’s an interruption engine designed to capture and redirect attention through notifications, social media, and an endless stream of dopamine-driven novelty. In a classroom, where even brief interruptions carry cognitive costs, that matters quite a lot. Students lose their place, their train of thought, their focus. Multiply that across an entire school day, and learning degrades.

The problem is not simply individual distraction. It is the erosion of the classroom conditions that make learning possible, even for students who aren’t on their phones in class. Schools depend on a basic level of order and predictability: students able to sustain attention, teachers able to direct it, and norms that hold without constant enforcement. Without these conditions, distractions are contagious. Phones, even when incorporated into instruction, dissolve that arrangement, replacing it with a steady stream of interruptions and a perpetual renegotiation of boundaries—“Put it away,” “Not now,” “Pay attention”—that teachers rarely win.

In this light, it becomes clear that the growing popularity of phone bans is about school culture and letting teachers teach, not a quick fix to underwhelming test scores.

Indeed, the most interesting finding in the NBER study was what happened not to test scores but to behavior. School suspensions increased at first, then stabilized. Student-reported feelings of well-being dipped, then improved.

That pattern suggests something deeper than a simple gain or loss. It suggests a reset. Remove phones, and the system does not immediately improve—it reconfigures. Students test boundaries, schools reassert them, and new norms (or restored old ones) take hold.

This is likely one reason schools with new phone bans sometimes report unexpected benefits. In one Texas district, library checkouts surged after students were required to lock away their phones during the school day. Students who once spent free moments scrolling began borrowing books instead. While we should avoid viewing the experience of a single district as universal, and examples like this shed no light on how students behave after school once their phones are returned, these anecdotes offer a glimpse into the benefits that could be derived from commonsense screens-in-schools policies.

The NBER study may not have shown the immediate, transformative academic impact that phone-ban advocates hoped for. But the speed with which schools have embraced these bans, and the changes in school culture that we’re seeing as a result, are genuinely impressive and worth celebrating. There is, however, a risk of overcorrecting. Enthusiasm for phone bans is already feeding a broader anti–screen movement—as if all screens are equally harmful and no screen-based technology can serve an educational purpose.

In several states, efforts that began with cell phone restrictions are expanding into broader limits on classroom technology. Missouri lawmakers recently approved legislation limiting screen time to 45 minutes a day and requiring most elementary school assignments to be completed with pencil and paper. Alabama, Tennessee, Utah, and Virginia are among at least a dozen states with legislation pending to restrict the use of technology in K–12 instruction and assessment. New advocacy organizations with names like Schools Beyond Screens and the Distraction-Free Schools Policy Project are positioning the issue as the next frontier in the parental-rights movement.

The impulse is understandable. But it risks treating all screens as if they pose the same problems and offer the same educational value, or lack thereof. That could be a mistake.

Schools were right to rediscover that attention matters and that not every instructional problem calls for a screen-based solution. They are also right to recognize that it’s not all about test scores, and the learning environment itself matters. Nevertheless, there is a danger in turning a necessary corrective into a general philosophy of technological abstinence. To a hammer, everything looks like a nail. To a school system newly disillusioned with phones and education technology, every screen can begin to look like a threat.

The useful question is not whether a tool, program, or lesson involves a screen but what it does to students’ attention and cognition. Does the screen have a legitimate instructional role to play, or is it a distraction or merely a mechanism to keep students “engaged”? Broad anti-screen sentiment misses this critical nuance.

To be clear, after years of advocates overpromising and under-delivering on educational technology, the backlash is understandable, even inevitable. For decades, schools were told that more devices, more apps, and more “engagement” would transform learning. In the main, they did not.

That’s why I’m a longtime skeptic of education technology. I never subscribed to fashionable claims about “digital natives”—the idea that children raised on technology somehow learn differently from previous generations of human beings. There is little evidence from cognitive science to support that notion. Human beings still learn the same way they always have: through attention, memory, practice, knowledge accumulation, and sustained effort.

Nor did classrooms filled with iPads, Chromebooks, and “interactive learning platforms” produce the education revolution their advocates promised. Too often, these devices merely digitized mediocre instruction (emphasizing activity-based learning and worship of student “engagement”) while adding new opportunities for distraction. As instructional coach Zach Groshell recently argued, the Chromebook often became less a learning tool than a classroom babysitter—a way to occupy students rather than teach them.

That said, substituting activity for learning didn’t start with computers. Yesterday it might have meant wheeling in a television cart or showing a filmstrip. Today it means handing every child a device and calling it personalized learning. The technology changes, but the temptation to confuse engagement, activity, and learning is a hardy perennial in American classrooms.

But it would be a mistake to conclude from this history that all screen-based technology is inherently and equally harmful. The problem was never technology per se. The problem was schools using technology to replace adult judgment, bypass effort, occupy or distract students, and dress up bad pedagogy.

Indeed, as Groshell observed, the common argument that “screens are not human” only takes us so far. “A book does not love the child reading it,” he writes. “A book does not kneel down to a child’s eye level.” Yet nobody argues books should therefore be banned. Why? Because, he argues, books do not replace human relationships; they extend them.

That distinction matters enormously. Technologies should be judged by whether they deepen human teaching or displace it, whether they strengthen cognition or circumvent it. Those questions are especially salient because of one emerging technology in particular: artificial intelligence. It’s becoming increasingly apparent that AI-enabled tools represent the greatest educational opportunity—and the steepest challenge—facing schools since the rise of mass public education itself.

The AI Classroom Crossroads

For decades, we have insisted that teachers practice “differentiated instruction” in mixed-ability classrooms, tailoring their lessons to each student’s needs, interests, and skill level. It is a lovely aspiration, but mostly that. In a classroom of 25 or 30 students with widely varying levels of knowledge and readiness, even the most skilled teachers struggle to provide the kind of individualized feedback and support we idealistically demand. Schools have spent decades pretending this can be done effectively; AI does it effortlessly.

I have seen AI systems give students detailed, actionable feedback on their writing—sometimes identifying weaknesses and patterns that I missed, even as a former teacher and professional writer. I have also seen early literacy tools use AI to guide beginning readers with real-time, individualized support, extending a teacher’s reach in ways that would otherwise be impractical or impossible. That is a meaningful breakthrough and a fundamental shift—not because AI promises to “engage” students (a claim we should now treat with deep suspicion) but because it expands teacher capacity.

This is what makes AI fundamentally different from most previous education technology, which was assumed to motivate students and make learning more appealing. AI may actually help teachers teach. Should these tools be banned simply because they involve a screen?

This is not to say that AI is a panacea. In solving one set of problems, it introduces an entirely new set of dangers. AI is dangerous when it substitutes for thinking, reading, writing, and student effort—and uniquely useful when it supports practice, feedback, tutoring, and differentiation.

The same technology that can provide individualized tutoring, immediate feedback, and targeted practice at scale can also weaken attention, bypass effort, simulate competence, and further atomize schooling.

Bad or inappropriate uses of AI are already obvious. A student using AI to generate an essay is outsourcing cognition itself. While phones distract students from learning, AI allows them to bypass learning entirely. Even more sinisterly, harmful AI use can be harder to detect than a student who is visibly distracted, which can complicate interventions and allow students to slip deeper into bad habits.

Students can now generate prose, summarize reading assignments, solve problems, and produce polished work with astonishing ease and minimal effort. Students have always found ways to cheat (by reading SparkNotes, for example, or copying a friend’s math homework), but AI presents a much greater temptation. It can produce work that appears coherent and intelligent even when the underlying thinking never occurred. It is nearly effortless on the student’s part and nearly impossible for educators to catch.

AI creates a temptation for not only students but also schools. Polished student work is reassuring. It suggests effective teaching, successful learning, and academic progress. When a student submits sophisticated essays and articulate responses, there may be little institutional incentive to probe deeply into whether the work reflects the student’s own thinking or a machine’s.

On the other hand, using AI for immediate corrective feedback or guided tutoring may be an excellent way for students to develop competence and knowledge. The question, then, is not whether schools should embrace or reject technology wholesale. It is whether they can use technology in a way that deepens human capacity rather than quietly eroding it.

For generations, schools largely assumed that the work students submitted reflected their own effort and understanding. But now, schools will have to rethink how they measure learning. That could mean greater reliance on in-class writing, oral examinations, and other forms of assessment that make student thinking visible. It will require teachers to distinguish between AI as a tutor and AI as a crutch—to help students understand when a question posed to an AI chatbot is supporting learning and when it is replacing it.

None of this will be easy. Schools will have to develop new norms, new routines, and perhaps even new forms of assessment capable of recognizing and rewarding genuine competence rather than merely polished output.

In the final analysis, there is an enormous tension at the heart of AI’s role in education. The same technology that may finally allow schools to provide individualized feedback and support at scale—and, bluntly, solve the problem of uneven instructional quality delivered by human beings whose own competence and motivation vary wildly—may also weaken the effort and struggle through which students develop genuine competence, intellect, and wisdom.

And the risks are unlikely to be evenly distributed. Students with strong background knowledge, verbal fluency, and intellectual discipline will likely use AI as leverage—a tool that extends their capabilities rather than masking deficiencies. Less prepared or motivated students may use it as a substitute for thinking, outsourcing judgment to systems they lack the knowledge to evaluate critically. In this way, AI may amplify the very inequalities it is uniquely well suited to solve.

Another risk is that schools may misread this moment in precisely the wrong way, using AI as an excuse to devalue academic knowledge even further than they already have. AI has made foundational knowledge more important than ever. Students cannot effectively evaluate AI-generated information if they lack the vocabulary, knowledge, and confidence to recognize errors, omissions, and nonsense.

The deepest challenge posed by AI, however, is not merely instructional. It is social and cultural. For decades, reformers have imagined increasingly individualized models of schooling: customized pathways, self-paced learning, and personalized instruction tailored precisely to each student. Artificial intelligence may finally make much of this technologically feasible. But our best schools are not mere content-delivery systems. They are social institutions charged with cultivating shared norms, shared knowledge, and common experiences.

In a commencement address earlier this year, I warned about the rise of what I called “schooling alone”—the gradual atomization of education into increasingly individualized experiences detached from common civic and cultural formation. AI could accelerate that trend dramatically. The same tools that make personalized learning newly possible may also weaken the shared experiences through which schools help form citizens, communities, and a common culture.

Balancing Backlash

The backlash against educational technology is, in many ways, justified. But schools would be wrong to conclude that every screen is a problem. If skepticism hardens into technological abstinence—if all screens come to be viewed as equally harmful—we risk oversimplifying an important conversation about AI in education. We could miss out on an opportunity to close stubborn and persistent gaps in student achievement born of unequal educational opportunities.

The challenge posed by AI is not whether schools should be “for” or “against” technology. It is whether we can learn to distinguish technologies that deepen competence from those that merely simulate it, technologies that extend human judgment from those that replace it, and technologies that strengthen the social and intellectual purposes of schooling from those that erode them.

Artificial intelligence may prove to be the greatest educational opportunity since the rise of mass schooling—and the greatest threat to competence, intellect, and shared culture since schools were created to cultivate them.

-30-

This article originally appeared at The American Enterprise.