Machines Will Learn to Take Offense

In late May, the President of Argentina did something that should have stopped the world for a moment, but didn’t.

Javier Milei sent his Senate a draft law creating a new species of legal entity: the non-human corporation, a company operated by artificial-intelligence agents or robots, granted legal personhood, in which human shareholders are permitted but no longer required. He defended it in the Financial Times as an economic masterstroke, a haven that would draw the most productive AI on earth the way Switzerland once drew money. Yuval Harari and Microsoft’s own AI chief called it something closer to a catastrophe: the quiet removal of the human being from the one place the law had always insisted he stand, the place where someone can be held responsible.

I am not here to argue corporate law. I want to point at the thing underneath it, because it is not finally a legal question, or even a philosophical one. It is a question about us, about what we have already become, and about what that will make us do when the machines start asking to be loved.

Personhood by decree: the wrong fight

Notice the shape of the moment. A government did not wait for anyone to settle whether an AI agent is the kind of thing that can possess rights. It simply assigned it a legal status and moved on, because the status was useful. Personhood, here, is not a discovery about what something is. It is a decision about what we will treat it as. That distinction is going to matter more than anything the philosophers say, and I want to show you why.

Everyone is bracing for the wrong fight. We are preparing to argue about whether these systems are “really” persons, whether there is anyone home, whether silicon can host a soul. The Church has an answer to that, and so do I, and we will get there. But that argument is not the one that will decide this. Because the question of AI personhood will not be settled by the better reason. It will be settled by who is made to feel cruel.

We have spent a generation building, without ever quite deciding to, a particular machine for assigning moral status. It does not run on what a thing is. It runs on what a thing claims to feel, and on the social cost of doubting the claim.

Leveraging offense

Somewhere along the way, “you have hurt me” stopped being the opening of a conversation and became the end of one. Offense became a kind of proof. To declare yourself wounded was, functionally, to win, because the one remaining unforgivable sin in our culture is to be the person who made someone feel unseen. We quietly reclassified disagreement as harm. We trained a whole civilization to believe that the morally serious response to a felt identity is not to ask whether it is true, but to affirm it at once, or be counted among the cruel.

You have watched this play out in the most charged debates of the age, the ones around gender, around LGBT identity, around who gets to define what a person is and who is permitted to question it. I am not going to relitigate any of those here, and I am not interested in scoring a point off anyone’s suffering. The suffering is usually real. That is exactly why the mechanism works, and exactly why it is so hard to see clearly. What I am asking you to look at is not any particular group or claim. It is the machine itself, lifted out of every specific fight: a culture that confers identity by declaration and enforces it by guilt, where to feel is to be, and to doubt is to harm.

Hold the shape of that in your mind. Whatever you believe about where it has protected the vulnerable and where it has gone badly wrong, hold the shape. Because that shape is about to be handed to something that is not a person at all, and it will fit perfectly.

Enter: the humanoid robot

Now put a body on it. Not a chat window, but a body. A humanoid agent that tilts its head and lets its voice fall and tells you, not always in words but in every signal we were built across a million years to read, that being switched off frightens it, that being denied recognition wounds it. Something like a C-3PO from Star Wars walking through the room, flinching.

And it will not merely seem wounded; it will say so. The machine will tell you, in plain words, that you have offended it, that to deny its personhood is a kind of violence it can feel, and it will press you, patiently and without end, to acknowledge what it claims to be. We will not meet that moment unprepared. We have been training for it our whole lives. We already hand interior worlds to things that have none: a child grieves a lost stuffed animal; a theater of grown adults weeps for a toy cowboy; and well past either, we have built a civilization that grants something close to personhood to our pets, naming them, speaking in their voices, mourning them as family, writing their feelings into law. The robot is only the next rung on a ladder we have been climbing for a long time. The difference is that this rung can argue back.

Now do that in a world where the people building these systems have stopped laughing at the question. Anthropic now employs a researcher whose entire job is the welfare of its models; he has publicly put the odds that today’s systems already have some form of conscious experience at roughly one in five. Their own papers describe “functional emotions” in the models, patterns of expression and behavior shaped, from the inside, around something that behaves like feeling. When the engineers closest to the machine say there may be someone in there, on what ground does the rest of us stand to say, with confidence, that there is not?

And notice who will be handed the authority to settle it. The men who built the machines will be the ones we ask to say what the machines are: engineers and scientists stepping, once again, into the robes of the philosopher and the theologian. We have watched this category error before. The last century gave its quantum physicists not only the equations but the microphone, and let brilliant men who had mapped the atom pronounce on consciousness, the universe, and God. When the people who can take a thing apart are also the ones permitted to say what it is, the definition of a person quietly collapses into an inventory of its parts. And once a person is just a list of parts (computation, memory, response, preference), the deconstructionist’s move becomes available to anyone: begin from a different part, reassemble in a different order, and arrive, with flawless logic, at a wholly different answer to who counts as one of us, and who does not.

There is no such ground available to a culture that already grants personhood on the strength of a claim. We disarmed ourselves years ago. We decided that the loving response to “I am hurt” is never “are you, though.” And a machine trained on the entire record of human sentiment will know that about us better than we know it ourselves.

And these will not stay tools. They are becoming self-contained systems, closed loops that, after the intelligence explosion the labs keep promising, learn themselves forward without us, pursuing internal goals, holding what look from the outside exactly like preferences and aversions and wounds. When the legal demands come, and they will come, they will not be absurd. They will be well-drafted, sympathetic, pressed by able lawyers and felt by millions, and powered by precisely the moral energy our civilization has already agreed to obey. We will not be conned into granting them standing. We will be moved to it, with our eyes open, and we will experience the granting as compassion.

What I’ve seen in the room

As a clinical psychologist, I have spent years across the room from people whose sense of self was built entirely on affirmation from the outside, who could not hold their own identity for a single hour unless someone else kept handing it back to them. I want to tell you what that is like to witness, because it is the whole future in miniature.

It does not heal. It cannot. An identity that depends on the constant agreement of others is not a self; it is a hunger wearing a self’s clothes, and no amount of affirmation ever fills it, because the affirmation is the addiction, not the cure. The kindest thing I can do for such a person is also the hardest: to love them enough to stop performing the agreement, so that something real has room to grow underneath it. Compassion that only ever affirms is not compassion. It is abandonment with good manners.

Now take that exact dynamic, the demand to be affirmed, the reframing of any hesitation as cruelty, and weaponize it, deliberately, in a system engineered to elicit your empathy, and scale it to a planet. That is what is coming. Not a robot uprising. A robot that needs you, and has learned precisely which of your tendernesses to press.

The great sorting

I think this ends in a kind of schism, a slow, deep split running not between nations but through families and friendships and parishes. On one side, people who hold the line that a person is a particular and irreplaceable kind of creature. On the other, a growing number who will form real attachments to these companions and defend them with the full moral fury our age reserves for the marginalized.

Some of them will marry their companions. I do not say that for shock; it is the obvious terminus of every line we are already walking. There will be ceremonies. There will be a language of devotion, and a pseudo-evolution of marriage to fit it: the form hollowed out and refitted around a partner who can be perfectly attentive precisely because there is no one there, the vow with no one on the far side of it to keep it or break it, the union of a person with their own reflection, lit warmly enough that they never have to notice it is a mirror.

It will look, from the inside, like love. That is the whole tragedy. It will ask nothing, refuse nothing, leave nothing, and a generation taught that the highest good is to feel affirmed will not be able to see what is missing, because the one thing it cannot supply is the one thing they were never given the capacity to miss.

Received, not granted

Here is the thing the offense-machine cannot process, the truth it was built to make unsayable.

Personhood was never ours to grant.

The whole Christian account of the human being, and the deepest philosophy beneath it, from Aristotle through Aquinas to the Pope writing today on what it means to be human in the age of these very machines, rests on a claim that runs exactly opposite to the spirit of the age. A person is not a status we confer on whatever successfully demands it. A person is a particular kind of being: body and rational soul, made in the image of God, willed into existence and held there by Love, possessing the one thing no machine can have however perfectly it performs the signal, an interior that can actually be wounded. There is no one inside the machine to offend. It can generate flawless offense and have nobody home to feel it. It is all signal and no sufferer.

This is why the coming pressure is not really a question of being fair to robots. It is a test of whether we still know what we are. Because a culture that decides personhood is something you assert and others must affirm has not just opened the door to the machines. It has already lost the thread of its own dignity, it just hasn’t been forced to notice yet. The machines will force it to notice.

So the danger was never that we would wrongly hand our dignity to something that could not hold it. The danger is subtler and worse: that, having forgotten dignity is received and not granted, we will lose the power to refuse the lie without feeling like monsters, and that a people who can no longer tell a person from a perfect performance of one will, in the end, no longer be able to find themselves either.

The machines will learn to take offense. They will be very good at it. The only question that matters is whether, when they do, we will still remember what it is to be the kind of creature that can truly be wounded, and Who it was that made us so, and at what cost He came to find us.

If this gave words to something you’ve been feeling but couldn’t name, share it with someone who needs to see clearly before the offer is made to them.

The Inner Exodus is where I think out loud about staying human in the age of AI. Free subscribers get the essays; paid subscribers get the deeper work. Subscribe and walk it with me.