Before either of us knew what an ornithologist was, my brother Ravi and I used to sneak out to the Sanjay Gandhi National Park at Borivali, two boys from a house at the edge of the park, climbing an old machan beside what had once been a pork factory. One evening a leopard walked out of the tree line into the clearing below us, unhurried, lit gold by the last of the light. Neither of us said a word. I remember that silence more than I remember the leopard.
Ravi kept that silence his whole life, though he learned to fill it with fact. He became an ornithologist, and for the better part of two decades I visited him in the field: the Andaman and Nicobar islands, the high Himalaya, the grasslands of central India where he tracked floricans through waist-high grass. Wherever I went with him, one bird kept returning to our conversations more than any other. The megapode.
Megapodes are, on paper, an unremarkable family of ground-dwelling birds found across the Indo-Pacific. What sets them apart is that their ancestors stopped incubating eggs with their own bodies. A megapode does not sit on a nest. It builds one — a mound of decaying leaf litter whose slow chemical rot generates heat, or a burrow warmed by sun or volcano — and walks away. The egg does the rest alone. When the chick hatches, weeks or months later, it digs itself out, and within a day it can fly, feed, and regulate its own body temperature. No parent teaches it anything, because by the time it needs to know anything, there is usually no parent left to ask.
Ravi spent nearly a decade with the Nicobar megapode, in conditions primitive enough that the islanders gave him a name of his own, Chidiya Babu, the bird man. He and his student built, over years of patient watching, what our family has always understood to be the first detailed record anywhere of how the species nests, defends its mound, and occasionally, remarkably, changes partners when the arrangement no longer serves either bird. It took years to know something that small and that certain.
What struck me, visiting him at the edge of those mounds, was the logic underneath the behaviour. A megapode gives up on parenting entirely, but only where the environment hands it a substitute: geothermal heat, or the reliable warmth of rot. Where no such substitute exists, no megapode has ever tried the trick. The behaviour did not arrive because it was clever. It arrived because the ground made it possible, and once possible, everything else — the giant yolk-rich egg, the chick born flight-ready — followed as consequence.
Institutions share that trait. They are not built from first principles, and they persist not because they are just or efficient, but because something in their environment — a resource, a threat, a coordination problem that would not solve itself — made a particular arrangement possible. What a bird cannot do, and an institution can, is rewrite the ground itself: a reform, a law, a revolution that changes the very conditions selecting for or against it. For most of institutional history, that rewriting has happened slowly enough that no single generation experiences it as change at all. What we are living through now looks different — several pressures arriving on top of each other rather than in sequence, chief among them the redistribution of cognitive labour by machines that did not exist twenty years ago. The smallest and oldest laboratory for what that does to an institution is the family raising a single child.
A human infant, measured against a megapode chick, arrives almost embarrassingly incomplete. It cannot regulate its own temperature, find food, or survive an hour unattended. One long-standing explanation, the obstetric dilemma, traces this to a bipedal pelvis narrow enough to walk on but able to pass only a skull of a certain size — so the infant is born early, and spends the next two decades finishing the job outside the egg, in the company of parents, grandparents, teachers, and whichever wider circle a society assembles for the purpose. Nearly everything we call family, school, and culture is, from this angle, an elaborate, distributed incubator: a mound built not of leaf litter but of relationship, doing slowly and expensively what the megapode’s ground does for free.
Every incubator, sooner or later, runs into a cheaper way of doing the same job. For the megapode, that was geothermal heat — free warmth that made the old method, sitting on the egg, an unnecessary expense. For a human childhood, the cheaper way may be arriving now, in the form of a machine that can compress into months what once took a lifetime of apprenticeship to transmit. If a machine can deliver, on demand, the procedural knowledge that once took a lifetime to earn, are we looking at the human equivalent of the five-times-larger egg — a childhood so front-loaded with capability that the long incubator of family and culture becomes optional rather than structural?
The maleo, a megapode cousin of Ravi’s Nicobar birds, lays the largest egg for its body size of any bird alive, over-provisioned with yolk, fuel for a body that must finish assembling itself before the shell cracks — an egg so rich it is also, not incidentally, worth stealing. What AI offers a human child is not fuel of that kind. It is closer to instinct: procedure, pattern, answer, delivered without the years of trial that would otherwise earn it. A sixteen-year-old with a competent AI tutor can now write working code for a well-defined task at a level that once took a working programmer years to reach. That does not make the teenager a professional; the judgment that comes from having done the work for a living is not the same as having produced one good instance of it. But a generation ago, even the instance was out of reach. Competence is what a person can do; formation is who they become in the years it takes to earn the doing. The chick that flies within hours of hatching flies the same flight every time, using a repertoire fixed at hatching and never revised. Human judgment — the kind Ravi spent years watching a pair of birds slowly develop, through trial, failure, and switched partners — is built from exactly the years AI now proposes to skip. AI can hand a child competence at a speed no apprenticeship ever matched. It is not yet clear it can hand a child formation at all.
A gun holds someone you love at the far end of it, in your own doorway, and the man behind it threatens to kill her unless you hand over money you don’t have. In the middle of the standoff, in some offhand remark or slip, you learn he is doing this to pay for his son’s medical treatment. Competence, however it is delivered, can tell you a great deal in that doorway: what tone of voice statistically de-escalates an armed and desperate man, what has worked across a thousand recorded standoffs, what the training manual says to do with your hands. It cannot tell you what to do with the fact that he is now, in the same breath, a threat and a father. That is not a gap in the data waiting to be filled by a bigger model. It is a genuinely undecidable weighing — mercy against risk, the life in front of you against the life he is trying to save, the small chance he is lying against the larger chance he is not — and no amount of pattern-matching resolves it, because there is no larger set of correct answers to match against. Formation does not resolve this either. There is no correct answer waiting under the panic to be found, only a choice that will be yours to make and yours to carry afterward, whichever way it goes. What formation gives you in that doorway is not the answer but the standing to choose one anyway, and to still recognize yourself in the choice once it’s made. Competence alone can produce an action here — a trained response, a practiced tone, a sequence of correct moves. But it is not obvious whose action that is. Formation is what makes the choice, however it comes out, unmistakably a person’s own, rather than a procedure that happened to run in a human body. Not every dilemma that looks unbearable has this shape. The classic runaway-wagon problem — divert the wagon from several strangers on the track onto one — looks agonizing and isn’t, really. Count the bodies, and any rule tells you which way to steer; a machine can already do this with more consistency than a person can. What makes the man in the doorway different is that nothing on either side of his sentence can be counted.
For most of history, the parent was a child’s first and last-resort teacher, the one who taught how to work, how to survive, how to read a room or a river. If competence no longer requires that teacher, the bond between parent and child loses one of its oldest justifications: that the child needs the parent to become anything at all. What’s left, once instruction is unbundled from love, is love itself, unobscured for the first time by function. Read hopefully, a parent freed from being a child’s sole gateway to the world might simply love more freely, less anxiously. Read as loss, something in the specific, halting texture of a father teaching a daughter to ride a bicycle, badly, patiently, more than once, is not obviously reproducible by a system that gets it right the first time.
Family was never only about affection either. It was also how a livelihood got passed down — the farm, the workshop, the trade, learned at a parent’s side because there was no other way to learn it. Care, culture, and competence travelled together, the way an incubator is not just a heat source but also a defense against predators and a gas-exchange system, folded together because building each separately cost more than nature already offered for free. The Nicobar megapode never had to bundle anything; it solved care, protection, and readiness all at once, at the level of the egg, and needed no family at all.
What AI threatens is not family itself but that bundle. If competence can be transmitted without years at a parent’s side, care, culture, and livelihood come apart for perhaps the first time in human history, and family is left holding only the first two. That doesn’t make family optional — no one chooses whose child they are — but it makes family’s hold on a person a matter of attachment rather than necessity: no longer the trade you had to learn because there was no other teacher, but the people you stayed close to because you wanted to.
Culture is the part most likely to travel with care rather than competence: a story told by an elder to a child who has no choice but to sit still and listen, values absorbed sideways, in the pauses between lessons, rather than delivered as content. This essay is one small instance of that channel. An AI-compressed childhood might deliver equal or greater competence and still fail to deliver this, because the transmission was never really about information. A society that raises its children on the megapode model risks producing citizens who are extraordinarily capable and oddly unrooted: fluent in everything, inheritors of nothing in particular.
The economic thread unravels most visibly, because it was always the most literal. Education systems, the timing of labour-market entry, pensions premised on generations overlapping in time — all of it built on the assumption that competence takes decades and a family to produce. Remove that assumption and these structures don’t erode gradually; they lose their foundation all at once, the way a bridge engineered for one load fails outright rather than sagging under a larger one. A twelve-year-old with front-loaded competence is either a prodigy or a labour-market entrant, and which one depends entirely on what the surrounding institutions decide to make of the fact. Most of them have no mechanism for deciding this at all.
Which leaves the oldest and most uncomfortable question: who gets the egg. A maleo egg, valuable enough to be worth stealing, is endangered today for exactly that reason. A human equivalent, competence transferred early and completely enough to change what a child can become, would be just as valuable, and just as unevenly distributed. The real danger is not a world in which every child hatches complete. It is a world in which some children do, and the rest are left, as before, to the slow mound, while the distance between the two groups compounds for the rest of both of their lives. Even the maleo cannot protect what its size makes valuable — a mound announces itself by the heat that makes it work, and guarding it costs more than a pair of birds can sustain across a season, which is why poaching, not predation, has emptied so many of its nesting grounds. Scholarships, public schooling, licensing exams open to whoever passes them, are all attempts to make competence something closer to a right than a theft. None has fully succeeded. An AI-compressed childhood works the same way the egg did: whoever gets it first pulls ahead, and everyone who arrives after them is running the same race from further back. If it spreads unevenly, it becomes only the latest and fastest version of that old theft — nothing taken from a nest this time, just handed to some children and quietly withheld from the rest, dressed up as nothing more than a gift freely given.
I was deep asleep when my father called at 3:30 in the morning on January 17, 2009, to tell me Ravi had been declared dead of a heart attack. There had been no illness to watch, no long decline to prepare for — one week he was in the field, and the next he was a voice on the phone telling me my brother wasn’t coming back. Ravi is gone now, and I no longer have anyone to stand with at the edge of a mound and ask which of us had the better door into the same room. But I keep returning, writing this, to the machan in Borivali, to the hour we spent in total silence watching an animal neither of us yet had words for. Nobody taught us that silence. We arrived at it the slow way, by sitting still together long enough for it to become the only correct response. I do not know what a child raised on the fast egg would have done in our place — possibly identified the leopard within seconds, and known something true about it that took Ravi another decade to learn. I want to believe that child loses nothing. AI is not going away, and wishing it were feels like wishing away weather; the honest question is not whether it should exist but what it actually costs. And when I hold that wanting up against what I watched Ravi build over a decade of trial and switched partners, it doesn’t survive contact. What formation gives a person is not information — it is the slowly earned sense of how to weigh one true thing against another when a situation hands you no rule for ranking them, only several things that matter at once. A child who has never had to sit inside that kind of not-knowing may arrive at every answer faster and still have no practiced way of choosing what to want when two things it values collide. Where formation would have built that judgment, in a child raised this way there may simply be nothing standing in its place — not ignorance, exactly, but an absence shaped like a path that was never walked. I only know that whatever Ravi and I built between us that evening — patience, kinship, a shared and unhurried way of looking at the world — took exactly as long as it took, and I am not certain that particular thing can be delivered any faster than it was delivered to us.
A note: this essay is personal, literary nonfiction rather than reportage. Ravi, our evening at Borivali, and his years of fieldwork in Andaman & Nicobar are drawn from family memory rather than from published sources. The account of his research — the partner-switching, the name islanders gave him — reflects how it has been told and remembered within the family, not a citation to his published work. The claims about megapode biology, the maleo’s egg, and the obstetric dilemma draw on general literature but are not individually footnoted here; a reader who wants the primary sources should treat them as a starting point for their own search rather than as settled fact.