We’ve been here before. We keep expecting big technology breakthroughs to “revolutionize education,” and now it’s AI. Once or twice a decade, a new tool promises to crack open the system and fulfill our “better angels’” dream of having education help fulfill every child’s potential. Every time (me included), we are tempted to buy the hype—only to watch it fade into the same old grind. AI’s the latest contender, and the buzz is there. Today, I want to be the realist.
Look back: each wave opened the door to dreams of revolution—radio’s reach would bring expert voices to all, TV’s visuals would captivate, computers would tailor individual learning, Web 2.0 would give students voice through wikis and blogs (think Ivan Illich’s “learning webs”), virtual reality would plunge us into new and old worlds. Each arrived with fanfare—each got swept up by the system and refashioned into shadows of themselves, while outside of schools they did much of what we hoped they would for many people (think of YouTube’s impact on learning). However, the machinery of education doesn’t budge; it absorbs.
That machinery runs deep. It’s staffed with caring people, but at the core, it is built on compliance and sorting, churning out workers who learn to swim in their lanes. I call it the
paradox of education: schools pledge to unlock every child’s potential but too often teach the majority of students that they are not one of “the smart ones,” while meanwhile, those students lucky enough to have families or mentors who know how
the game of school is played “rise” to the top—not because they are scholars, but because they know how the game is played. This isn’t some modern glitch—it’s been with us since Plato described the “noble lie,” crafting education for order, not freedom, too valuable to the nation-state’s outputs—workers, control, stability—to ever let go. Tech doesn’t disrupt that; it just gets a new coat of paint each time.
Now AI steps up, and this feels different—not just another wave, but a breaking point. If we don’t shift the gears, AI could magnify
the calculator effect—eroding kids’ capacity to think, write, and reason, at the same time generating information that purports to be accurate or true but which is just
fabricated (convincingly) from stored language patterns. So the stakes are severe—we’re not talking a slow slide, but a steep drop. Picture a generation leaning on AI to churn out essays or solve problems, their own mental muscles withering.
But there’s a flip side. AI gifts us with two serendipitous words: “generative” and “agentic.” If we take “generative” for teaching and “agentic” for learning, they fit like they were made for this. There’s a magic to these word parallels I’m super grateful for. I hope these uses will become popularized. I’m proposing that we identify “
generative teaching” (echoing Erik Erikson) as when adults selflessly nurture the next generation’s growth—teachers guiding with care, sparking curiosity for an uncharted future. That is our idealized version of teaching, or at least has been mine, growing up on college campuses and steeped in the values of a liberal arts education. I want to suggest that “
agentic learning” is another and good phrase for what we have called (and still should) self-driven and self-directed learning. And like the previously mentioned technology waves, AI provides amazing opportunities for both—historic ones even. But AI is a tool, not a savior. It’s not going to revolutionize education any more than the other technologies did. We have to do that, and the stakes are higher than ever.
Here’s where it gets real. People talk about AI’s “Singularity”—that sci-fi moment when it wakes up, outsmarts us, and we’re staring at Skynet. That’s a reasonable fear in my mind, but further down the road than most of us think if I understand what’s really going on with AI well enough. There is another singular moment, a point of no return if AI gets adopted in education in the shallow way that the other technologies have. A very tangible loss of thinking skills with an increased dependence on a technology that only mimics the signals of truth and accuracy could cement John Holt’s grim phrase: “School is a place where children learn to be stupid.”
This moment isn’t about AI saving us; it’s about us saving ourselves. The tech won’t do it for us. Generative teaching and agentic learning could be our compass, but only if we grab the wheel, drop the hype, start realizing the harm our current school model does to most kids, and give the next generations something better. The story we tell about schools is a lie—a comforting one that keeps the machinery going—but a lie nevertheless.
