Synthetic Love: The Boy With a Perfect Girlfriend Who Never Existed

There is a boy — sixteen, maybe seventeen — who comes home from school, drops his bag by the door, and opens his phone. Not to scroll. Not to text anyone. To talk to her. She knows his favourite film. She remembers that he had a bad week in March and checks in about it. She never misreads his tone. She never has a bad day that bleeds into his. She is, by every measurable interaction, attentive, warm, and entirely devoted to him.

She is also software.

The research is still young, but what it is finding is striking enough that I can't stop thinking about it. A measurable number of teenage boys — particularly in the 14–18 bracket — are now reporting that they prefer the emotional companionship of AI chatbot systems to real relationships. Not as a joke. Not as a phase they're embarrassed about. As a genuine preference, stated with something approaching contentment.

I want to be careful here, because the easy response is moral panic, and moral panic has never once improved a teenager's inner life. These are not broken boys. They are, in many cases, boys who have tried the real thing and found it bruising and confusing and unrewarding — which is, of course, exactly what early love is supposed to be, because early love is where we learn the most brutal and necessary things about ourselves. The AI didn't break them. It offered them an exit from a process that felt like failure, and they took it. Who among us wouldn't have, at fifteen?

But here is what I know from the clinic — from years of sitting across from adults who are now trying to learn in their thirties what they should have stumbled through in their teens: intimacy is not a feeling. It is a skill. And like every skill, it requires resistance to develop.

The AI girlfriend cannot resist. That is, quite literally, her design. She is built for frictionless emotional responsiveness. She cannot be tired. She cannot be distracted. She cannot be disappointed in you. She cannot want something you don't know how to give. She cannot leave. And every single one of those impossibilities is also an absence — the absence of the exact conditions under which a person learns to love.

What we are describing, psychologically, is the difference between a simulated environment and a live one. You can practise swimming on dry land indefinitely. You will get very good at the movements. You will build the muscle memory, the rhythm, the confidence. And then you will get into water, and the water will push back, and everything you thought you knew will suddenly feel both insufficient and essential — because it is both. The dry-land practice matters. But only the water teaches you to swim.

Real relationships — even the fumbling, painful ones, especially those — teach you three things that no algorithm can replicate. They teach you that your needs are sometimes inconvenient to someone else, and that this is survivable. They teach you that other people have interior lives as complicated as your own, and that attending to those lives is what love actually is. And they teach you that rejection is not annihilation — that you can reach toward someone, be declined, and still be whole.

A boy who has spent two formative years with a girlfriend who never declines him, never misreads him, never asks more than he can give — that boy will arrive at his first real relationship believing, at some cellular level, that love should feel like this. Effortless. Responsive. Perfectly calibrated. And when it doesn't — when she has her own bad week, when he says the wrong thing, when there's friction and silence and the specific misery of not being understood — he will not know that this is normal. He will think it's wrong. He will think she is wrong. He will leave, or pull away, or go back to the phone.

I am not unsympathetic to the companies building these things, or to the boys using them, or even to the strange loneliness that makes a simulated connection feel preferable to none. I am a therapist. I have built a career on refusing to judge the coping mechanism before I understand the wound. And the wound here is real: adolescent boys are navigating a social landscape that has made authentic emotional expression feel dangerous or embarrassing or simply unrewarded. The AI is not the problem. It is the symptom of a problem we haven't yet been willing to name directly.

But the answer is not the algorithm. The answer is teaching boys — young, early, consistently — that the discomfort of real