Shekhar Natarajan has spent the better part of two years telling rooms full of policymakers and executives a story they think they already understand. The streetlight in South Central India where he studied as a child. The thirty rupees his mother received for pawning her wedding ring. The thirty-four dollars he carried when he landed in the United States. The Fortune 500 boardrooms - Walmart, Coca-Cola, Disney, PepsiCo - that came after. The seventy-plus patents. Audiences hear the arc and nod, because it is the kind of arc people have learned to associate with moral seriousness. The man asking the hard questions about artificial intelligence, the thinking goes, must have been forged by the distance between those two ends of his life.
Over the past two years, that thinking has produced its own consequences. Natarajan has been recognized at the World Economic Forum in Davos, at the AI Summit, and at Oxford University - a sequence of platforms that, taken together, has positioned him as one of the leading voices in AI ethics working today. He is no longer a supply-chain technologist with an unusual side argument. He is, increasingly, the person the rooms are organized around when the conversation turns from capability to conscience.
Lately, in those same rooms, Natarajan has begun to push back on the interpretation that travels with him. The arc, he argues, is not what taught him. It is not the source of the ethics he is now trying to embed inside Angelic Intelligence, the system his Dublin, California-based company Orchestro.AI has been building since 2023. The full crossing - from poverty to plenty, from obscurity to the C-suite - was not the curriculum. It was, at most, the credential.
The teacher, in his telling, is a woman who never crossed at all.
The argument against the arc
The dominant assumption in moral philosophy, and increasingly in the ethics of machine learning, is that wisdom is comparative. One cannot understand sufficiency without scarcity, the reasoning goes, or power without powerlessness. Moral judgment, on this view, is the residue of contrast - and contrast requires traversal. The immigrant who became an executive is wise because he remembers the streetlight. The executive who never knew the streetlight is suspect. The child still studying beneath one cannot yet teach anyone anything.
Natarajan thinks this is wrong. Not in part - in whole.
His central counter-example is his mother. She never crossed. She stood outside a headmaster's office for a year to secure his admission to school. She pawned her wedding ring for thirty rupees to pay his fees. She raised five children on an income that would not have fed one, and she did not, at any point in her life, see the other side of the journey she set him on. She did not sit in a boardroom. She did not hold a patent. She had one shore, and she had it her whole life. Every virtue Natarajan now says he is trying to encode into a machine - compassion, dignity, truth-telling, restraint - he traces, when pressed, to her. Not to the contrast. To her.
What one side contains
The claim, stated plainly, is this: a human life lived fully on a single side of any journey already contains every moral primitive a machine would need to learn. Compassion does not require having been comfortable in order to recognize suffering. Truth-telling does not require having lied at scale and recovered. Dignity does not require having been humiliated and restored. Each of these, Natarajan argues, can be learned and held in full from one position, by one person, across one ordinary life.
The idea that ethics requires a full traversal - poverty to wealth, weakness to power, obscurity to recognition - is, in his view, a Silicon Valley idea dressed up as a universal one. It is the founder's myth applied to morality. It says that only those who have made the crossing are qualified to draw the map. But the people he cites as having taught him most about being human - his mother, the teachers in his village, the neighbors who fed him when his family could not - never made any crossing at all. They lived in one place. They suffered and endured and refused cruelty in one place. What they knew, he insists, was complete.
Why the architecture changes
This is not, in Natarajan's hands, a sentimental point. It has direct consequences for how he believes AI systems should be built.
If ethics requires the full arc - if wisdom emerges only from contrast - then a moral machine must, by implication, be trained on opposites. It must be fed harm and benefit, falsehood and truth, exploitation and care, in the hope that some kind of judgment will emerge from the gradient between them. This is, broadly speaking, the regnant paradigm in AI alignment: reinforcement learning from human feedback, constitutional methods, red-teaming, content filters. The system is exposed to the full distance of human behavior and trusted to navigate it. Critics, Natarajan among them, argue that the resulting models feel hollow precisely because they have learned the shape of ethics by triangulating between its violations rather than by being grounded in any particular vision of the good.
If, instead, one side of the journey is enough - if a single life lived with integrity already contains the whole moral grammar - the architecture changes. The machine no longer needs to traverse the full distance from cruelty to care in order to choose care. It can be built from one side. Its foundation can be the orientation of the woman who stood outside the headmaster's office, and that orientation, Natarajan argues, will generalize. The system does not need to know what corruption feels like in order to refuse it. It needs to know, structurally, what loyalty feels like. The refusal follows.
This is the argument behind a line he has repeated in venues from New Delhi to Davos: virtue cannot be a guardrail. A guardrail, he says, assumes the machine will drift toward harm by default and must be fenced away from it - the architecture of a system trained on both shores and trusted to stay on neither. A machine built from one side, in his framing, does not drift. The fence is unnecessary because the gravity is correct.
The objection
The most common objection - one Natarajan acknowledges he hears from researchers he respects - is that without exposure to contrast, neither a person nor a machine can recognize harm when it appears. A system that has only seen kindness, the argument runs, will be naive before cruelty. Some traversal of the moral terrain is necessary for the moral muscle to function at all.
His response is that recognition does not require enactment, and exposure does not require traversal. His mother, he points out, recognized injustice without ever having committed it. She recognized greed without having been greedy, and cruelty without ever having been cruel. The asymmetry of moral knowledge - the fact that a person can know harm intimately without ever inflicting or sustaining it - is, he notes, one of the oldest observations in ethical thought, present in the Sanskrit, Greek, Arabic, and Confucian traditions that inform the twenty-seven Digital Angels at the operational core of his system. Compassion, in each of those traditions, is not the residue of having crossed from suffering into ease. It is the steady recognition, from wherever one stands, that another being's suffering matters. A machine, he believes, can be built that way.
What the arc is actually for
If the arc is not the teacher, why does Natarajan keep telling the story? The streetlight, the ring, the thirty-four dollars - these appear in nearly every keynote, every long-form profile, every investor deck. He is candid about the function. The arc is not what taught him; it is what gives him standing in the rooms he now occupies. It is a credential, not a curriculum. He acknowledges the credential is useful, and he is not interested in pretending otherwise. But he resists the conclusion most listeners draw from it - that the journey was the source of the wisdom.
Every founder telling a journey story is, in his view, asking the audience to believe that the journey was the teacher. He is asking them to believe something less flattering and, he thinks, more accurate: that the teacher was already there, on one shore, before the journey began. The journey only carried the lesson into rooms where it could be heard. It did not generate the lesson. It transported it.
And if that is true of a person, he argues, it can be true of a machine. A system does not need to be dragged across the full distance of human moral experience to be made human. It needs to be rooted, at its foundation, in the orientation of a single life lived honestly on a single side. That, he says, is enough. It was enough for his mother, who taught him everything, and who never went anywhere.
A necessary caveat
Natarajan is aware that the argument, taken loosely, can be used to let the comfortable off the hook. If one side is enough, the wealthy can claim moral standing without having known scarcity, and the powerful can claim wisdom without having known constraint. Pressed on this, he draws a sharper line. One side is enough, he says, only if the side is lived. Comfort that has not been examined is not a side of the journey. It is sleep. The mother who stood outside the headmaster's office had lived her side. The executive who has never paused to consider the people moving through his warehouse has not lived his.
The question, in his framing, is not which shore a person stands on. It is whether they have stood there awake.
If they have, he argues, they already contain what the machine needs to learn. And if they have not, no crossing to the other side will give it to them.
