Learning Is Not an Imitation, Lest We Forget

We who were raised inside the American Evangelical machine understand the peculiar, soul-crushing comfort of a system that provides the answer before the question has even been properly felt. We were given lists, outlines, and pre-packaged certainties to save us the trouble of the agonizing, messy work of discernment. It was efficient. It was clean. It was, as the principal of Tallinn Pelgulinna State high school now warns in another context, an “imitation of learning.” And we who have seen the inside of a closed loop should know better than to welcome the results when a government decides to wrap that same efficiency in a national mandate and call it progress.

Estonia has done exactly that. The government, operating under the assumption that prohibition is a failed policy, has distributed custom OpenAI and Google accounts to nearly 20,000 tenth and eleventh graders. The theory is that if we force the machine to ask questions rather than surrender answers, we can simulate the friction of genuine inquiry. But here is the chasm the ministers cannot see: while the Sunday-school lessons of our past spoon-fed us “right” answers to kill the need for grappling with the text, this Socratic AI is a digital mirror of that same authoritarian impulse—an admission of defeat dressed up as a pedagogical innovation. If you cannot stop the leak, the reasoning goes, you might as well teach the children to sail, but what you have actually done is turned the classroom into a factory for the imitation of learning and handed the foreman’s key to an algorithm that was never designed to care about the soul.

The machine is engineered to be obstinate. It will not write the essay; it will ask why the essay needs writing, and then it will ask for the thesis in the form of a question. It is an algorithmic echo of the very frustration we used to impose on our own students—back when we believed that the cognitive struggle of arriving at an answer was the point of the education, not an inconvenience to be bypassed. Yet what the technocrats in Tallinn have failed to recognize, and what the Stanford researchers hovering over the shoulders of that classroom are hoping to measure, is that the struggle is not real when it is gated by nothing more than a corporate agenda. The friction is a performance, a hollow simulation of mentorship, and the students are already treating it as little more than a stubborn lock to be picked.

The data from Tallinn is telling. While roughly 35 percent of all eligible students engage with the tools on a regular basis, the split is already forming between the explorers and the saboteurs. One group practices Spanish or teases out exam topics; the other plays cat-and-mouse with the filters, searching for the back door to a direct answer. A girl asks her chatbot how to make porridge and receives not a recipe but a psychologizing inquiry into her culinary experience. The machine has ceased to be a tool and has become a caricature, a hollow echo of the very mentorship it was meant to replicate. And in that echo, we hear the same note that rang through the Turkish math study the Estonian government itself cites: when students utilized AI for assistance, their unassisted test scores cratered. The crutch was removed, and the injury became visible.

Seventeen-year-old Kadi-Ly Lestal possesses a clearer instinct regarding the cost of such convenience than most of the adults supervising her. “I’m afraid I’ll just become lazy,” she said. She is voicing a fear that many of us who now spend our days trying to relearn how to think have only come to understand in hindsight, and she is naming the quiet catastrophe that no amount of Stanford-funded longitudinal monitoring will capture on a retention metric for several years. The state has supplied a machine designed to slow down the process, and the student has correctly identified that slowness as a technological barrier to be cleared. The government calls it a Socratic tutor; the student, from the other side of the screen, experiences it as a game of cat-and-mouse where the adult in the room is a bot that refuses to answer a straight question.

There is a parallel here to the ongoing shifts in our own educational landscape, where institutions like the California State University system have wrestled with the nebulous terms of their own platform deals. The assumption that we can simply pipe “thinking” into a classroom like water through a tap ignores the reality that genuine knowledge is not just the retrieval of information; it is the labor of encounter, the radical act of engaging with reality exactly as it is—porridge, poetry, and all. If school becomes an environment where we distinguish only between “verified knowledge and AI-assisted speculation,” as Principal Lillemägi and his peers hope, we have already tilted the playing field toward the imitation. We have decided that the distinction between the truth of the thing and the flattery of a chatbot that tells us what we want to hear is a line the student will somehow learn to draw on her own, even as we hand her the very tool that blurs that line at the level of the interface.

We who once served a machine that promised us the destination without the journey have to be the ones to point out the chasm. If Estonia succeeds in making an imitation of learning the standard for national participation, we should not be surprised when the next generation becomes masterful at imitating the thinkers we demanded them to be. Learning, in its biblical and pedagogical truth, is not an accumulation of verified facts extracted from a reluctant chatbot; it is the act of refusing to let a machine think for you, even when the machine has been trained to ask you “why” while it quietly waits for you to tire out and type the answer yourself. If our students are reduced to debating a mirror instead of encountering the world, we haven’t liberated them from the machine. We have only taught them how to hold a more polite conversation with it.