You Cannot Prepare Children for the Future by Banning the Future

On September 2, 2026, New York City Mayor Zohran Mamdani and Schools Chancellor Kamar Samuels announced a one-year moratorium on student-facing generative AI for students in grades 2-K through 8th, affecting nearly 600,000 students, roughly two-thirds of the district's total enrollment. High school students will receive twice-yearly AI literacy instruction and limited, supervised access through a small number of pilot programs. This is not a permanent prohibition on artificial intelligence. It is, by the city's own description, the most expansive AI restriction of its kind in the country, and it raises a question every school should be thinking through carefully.

The administration's stated concerns are not frivolous. Mayor Mamdani has argued that children need interaction with teachers and classmates, creativity, discussion, and the chance to work through difficult problems rather than immediately outsourcing them to a machine. Those are the right priorities. Where the policy runs into trouble is the assumption that keeping students away from AI entirely is the best way to protect them. The controversy actually exposes a much deeper problem in American education, one that predates AI by decades.

Rote Learning Is Not the Education Worth Protecting

Much of traditional education has been organized around rote learning: memorizing facts, repeating procedures, completing worksheets, and reproducing the answer a teacher or textbook expects. A student can become extremely proficient at this kind of schooling while understanding surprisingly little about why anything actually works.

That is precisely the kind of education The Barrett School is built to move away from. Successfully reproducing information is not the same thing as understanding it, and memorization should never be confused with intellectual mastery. If a student knows the correct answer but cannot explain why it is correct, apply the concept in a new situation, or question an assumption behind it, the school has taught performance rather than understanding.

Consider the common analogy used in debates over educational technology: that children must memorize multiplication tables before being allowed to use a calculator. There is real value in mathematical automaticity, but knowing that seven times eight equals fifty-six is not the same as understanding multiplication. A child should be able to see seven groups of eight, recognize multiplication through arrays and repeated addition, and understand why multiplication and division are related. The memorized fact should be a convenient consequence of that understanding, not a substitute for it. Otherwise the child has only memorized an output.

That distinction matters enormously for AI, because generative AI is extremely good at producing outputs. It can write the essay, answer the vocabulary questions, summarize the chapter, and reproduce enormous quantities of information almost instantly. If those activities are what most schools consider education, AI genuinely is a threat to that model. But that should prompt a different question: if a machine can complete an assignment perfectly without understanding anything, was the assignment ever measuring understanding in the first place?

AI Can Force Assignments to Get Better

For generations, schools could reasonably equate the production of work with the process of learning, because there was no real alternative. A three-page report on Ancient Rome implied the student had spent time researching Rome. Several paragraphs comparing two novels implied the student had actually read them.

Generative AI breaks that assumption entirely. A machine can now produce the report without reading anything, understanding anything, or thinking about the question in any human sense. The obvious response is to ban the machine.

A more useful response is to redesign the assignment so that producing words is no longer mistaken for demonstrating understanding: ask a student to defend an argument orally, compare contradictory primary sources and explain which is more trustworthy, build a model, design an experiment, or identify the weaknesses in an AI-generated answer. Ask not merely what a historical figure believed, but whether the argument actually holds up, where it fails, and what a student would say in response.

That is considerably harder for both student and teacher to do. It is also considerably closer to real education.

Productive Struggle Is Not the Same as Unproductive Difficulty

The moratorium's emphasis on intellectual struggle contains an important insight. Students should encounter uncertainty, frustration, and questions they cannot immediately answer; genuine intellectual development requires the mind to reorganize itself around something difficult.

But difficulty alone does not produce learning. Copying a vocabulary word fifty times is tedious, not profound. Making children manually repeat procedures they do not understand can create the appearance of rigor without the intellectual transformation real learning requires. Productive struggle has a purpose: it happens when a student is genuinely trying to understand why something works, testing possibilities, or revising an idea. The goal was never difficulty for its own sake. It was always understanding.

This is exactly where technology has historically helped rather than hurt. A calculator can handle arithmetic while an engineering student reasons about the physical system being modeled. A word processor can handle formatting while a writer focuses on argument and structure. AI can potentially do something similar, if the question schools ask is which cognitive work should stay with the student and which mechanical work can reasonably be delegated. That is a question for good teachers and thoughtfully designed assignments, not for blanket prohibition.

Teaching Students to Distrust AI, Intelligently

One of the more counterintuitive ideas in good AI education is that allowing students to use AI does not mean teaching them to trust it. A strong AI curriculum should do close to the opposite. Give a middle schooler an AI-generated explanation of a historical event containing several subtle errors, and ask them to find them. Require locating primary sources, distinguishing claims from evidence, and explaining exactly where the AI went wrong. Suddenly AI has become the object of scrutiny rather than a source of authority.

Ask several AI systems the same question and compare the answers. What assumptions does each make? What evidence is missing? Which response actually holds up? A classroom can turn AI from an answer-dispensing machine into an intellectual sparring partner, and the lesson worth teaching is never "the computer knows." It is "show me how you know."

UNESCO's AI Competency Framework for Students takes exactly this broader view, built around four dimensions: a human-centered mindset, the ethics of AI, foundational AI knowledge, and AI system design, all aimed at producing students who can examine AI critically rather than simply consume what it produces. That is a far more ambitious goal than keeping the technology out of the building entirely.

Schools Have Been Here Before

The pattern is familiar. Calculators once raised widespread fears that students would stop learning fundamental mathematics. By the late 1970s, educators had largely settled on a more sophisticated position: students should understand basic operations first, then use calculators for estimation, comparison, and problem-solving, rather than treating the calculator as either forbidden or a replacement for mathematical thinking.

Computers generated similar anxiety in their own era. During the early 1980s, "computer literacy" quickly became an educational priority as schools recognized students were entering a world in which computers would shape employment, communication, and everyday life, with computer access expanding rapidly through the rest of the decade. Someone could have argued that children did not need computers because handwriting, books, and arithmetic already existed. Most schools instead concluded that students needed both traditional skills and the ability to understand the emerging tools around them. AI belongs in that same tradition, though its power calls for correspondingly stronger safeguards.

A Moratorium Can Widen the Gap It Means to Close

There is also a real equity problem with removing emerging technology from public education specifically. Families with resources will not stop their children from learning AI because a public school system restricts it. They can purchase subscriptions, hire tutors, enroll children in camps, and provide access through their own professional environments. Families without those resources have no comparable option.

That means a public-school moratorium can produce the opposite of its intention: rather than protecting children equally, it can make AI literacy something acquired privately by families who already have the knowledge and means to provide it. Public education should be closing that gap, not widening it. UNESCO has repeatedly flagged genuine risks around AI in education, including inequality, privacy, and safety, but its own response centers on human-centered, equitable integration rather than the assumption that schools can simply turn away from the technology. That distinction matters.

AI Does Not Replace the Teacher. It Makes the Teacher More Important

None of this argues for unrestricted access to commercial chatbots for young children. Age limits, privacy protections, supervision, and clear boundaries around AI use are all reasonable. There are legitimate reasons to require some work be done independently, so a teacher can see what a student actually understands.

But those are arguments for good pedagogy and responsible policy, not arguments against AI literacy itself. A competent teacher in the room can stop a student and ask why they trust a particular answer, demonstrate what a hallucination looks like, or simply close the laptops because the next problem needs to be solved without one. A blanket moratorium cannot do any of that. A teacher can.

The Real Threat Is Passive Thinking, Which Predates AI

A generation of children who ask a machine every time they encounter uncertainty would be a genuine loss. But rote education has already spent generations teaching a different form of passive thinking: a child is given information, memorizes it, repeats it on a test, and moves on, often without ever challenging the idea or understanding why it mattered.

AI is an opportunity to confront that older problem rather than preserve it. The right standard is neither "AI knows everything" nor "AI must be kept away from children." It is a more demanding proposition: a student is responsible for understanding what they claim to know, regardless of which tools helped them get there. That standard is harder. It is also better education.

What This Means for How The Barrett School Approaches AI

The debate playing out in New York points to a genuine choice every school faces: treat AI as a threat to be walled off, or treat it as a technology students need to learn to use, question, and understand. The Barrett School has built its STEM and technology curriculum around the second approach from the Early School years forward, on the premise that the students least prepared for an AI-shaped world are the ones who were never taught how it works.

The article on STEM and AI education in Destin, Florida covers how this looks in practice across every grade, and the article on what makes a private school STEM program actually strong covers how to tell a genuine program from a marketed one.

Schedule a campus visit to see how technology and critical thinking are woven together in daily classroom instruction at The Barrett School. Tours run Monday through Friday between 9:00 AM and 3:00 PM. The admissions overview covers enrollment steps, and the team is available at (850) 353-2153 or info@thebarrettschool.org.