A child today can produce a confident answer to almost any question before they have formed a single thought of their own. That is the quiet danger we have walked into. For every generation before this one, learning demanded friction — we searched, we compared, we decided what to trust, and somewhere in that struggle the habit of thinking was built. Remove the friction, and we risk raising a generation that can retrieve everything and reason about nothing.Artificial intelligenceBut this is not a story of despair, and we should refuse to tell it as one. The most useful frame I have encountered comes from Harvard, where educators describe what they call Hybrid Intelligence — AI not as an oracle to be obeyed, but as a thinking partner to be questioned. The machine and the human each bring what the other lacks: The machine brings pattern and scale, the human brings judgment and conscience. Our task is not to keep children away from AI, nor to surrender them to it. It is to teach them to think with it — and, just as crucially, to think beyond it.That distinction changes what a classroom must do. Four shifts matter most, and not one of them is technical.The first is a shift from searching to researching. For a century, school rewarded the child who could find the right answer fastest. AI has made searching free. The scarce skill now is researching and examining — asking whether an answer is sound, where its evidence comes from, and what it quietly leaves out. We must stop grading retrieval and start grading discernment.The second is to cultivate the discipline of healthy scepticism. A child who accepts a polished AI response without a flicker of doubt has not been failed by their own curiosity; they have been failed by a curriculum that never taught them to interrogate. From the age of eight or nine, children can learn to ask three questions of any output: Who made this? How do we know it is true? What might it be getting wrong? Asked often enough, those questions stop being exercises and become instincts.The third is to prize original thinking and creative intelligence. AI is extraordinary at recognising patterns and synthesising what already exists; that is also its ceiling. It remixes the known. It cannot wonder, cannot want, cannot feel the stakes of what it says. So, the capacities that remain ours — empathy, ethical reasoning, original creative expression, the idea that begins where the training data ends — are no longer soft extras at the edge of the timetable. They are the core curriculum of a human being. To think with AI is one skill; to think beyond it, to make something the machine could not have predicted, is the rarer and more precious one.The fourth, and perhaps the hardest, is metacognition — teaching children to think about their own thinking. A child needs to notice when they are offloading too much of their mental work to a tool, the way we once noticed our handwriting weakening when we stopped writing letters. The exercise is simple and revealing. Have a child write an essay, or make a drawing, entirely on their own. Then have AI produce its version of the same task, and sit beside them as they study the difference. Where is the voice? Where is the effort? What did the machine smooth away that was actually theirs? A child who can name that difference has learned something no algorithm can teach them — the worth of their own mind.India is not starting from nothing here. NEP 2020 and the National Curriculum Framework for School Education both name AI literacy as a priority, and the policy intent is real. Now we need a curriculum that is genuinely age-differentiated, teachers prepared to lead a conversation about why a machine can be confidently wrong, and assessments that finally measure reasoning rather than recall. This is an adaptive challenge, not a technical one — it asks us to change what we value, not merely what we buy.None of this means turning away from the technology. It means refusing to let our children become passive consumers of it. And children learn to question by watching adults question. The teacher and the parent who thinks aloud — who wonders openly, tests a claim, admits what she does not yet know, and unpicks an AI answer line by line, asking why — is modelling the very inquiry we want children to own. That is the most important work in education today: building the expectation that every output deserves scrutiny, and that the final judgment always belongs to the human being.The question before us was never whether children will use these tools. They already do. The question is whether we will teach them to amplify their thinking or to suppress it — especially when the tools are wrong, incomplete, or built on assumptions no one has asked them to examine. That is the work. And the journey begins with us.(The views expressed are personal)This article is authored by Manit Jain, founder, Humain Learning and co-founder, Heritage Schools.