Sex Robots, Reconsidered
Everyone knows what the words mean. Almost no one agrees on what the thing would be.
Say “sex robot” and the object appears before the argument does: a human-shaped appliance, permanently available, built to turn desire into a command and a body into a service.
The judgment arrives almost as quickly. Sad. Creepy. Degrading. Perhaps dangerous. None of those reactions is frivolous. They come from real fears: that intimacy can be bought without responsibility, that a body can be designed without a will, that loneliness can become a market instead of a condition we try to change.
But the phrase hides as much as it reveals. It joins two words we think we understand—sex and robot—and produces an object no one can quite define. Is it a toy with a face? A service with a body? A companion that can touch you? A new kind of relationship?
We have judged the category before agreeing on what belongs inside it.
One phrase, three different machines
“Sex robot” currently collapses at least three very different ideas into one image.
The first is an advanced sex toy: a device that can move, sense and respond to the body. It may look human, but it makes no claim to be a companion. It is still an object someone uses.
The second is an embodied service: a machine that learns preferences, performs a personality and adapts itself around one user. It can create the feeling of interaction while remaining organized around a single purpose—satisfaction.
The third is more difficult to name. It is a synthetic companion whose body is only one part of an ongoing relationship: something that remembers, develops continuity and remains present before and after sex. Here, sex is not the whole product. It is one of the places where the relationship becomes physical.
These machines are not interchangeable. The first can be judged as a device. The second raises questions about service, simulation and dependency. The third asks whether a product can become a participant in someone’s intimate life.
Much of the argument around sex robots fails because one person is condemning the third machine while another is defending the first. Reconsideration has to begin by asking what object we are actually talking about.
Sex is not the simple part
Sex looks easy to automate because it has visible mechanics. Bodies move. Nerves respond. A machine can reproduce motion and create sensation. None of that tells us whether someone feels desired.
Sex can exist without intimacy, and intimacy can exist without sex. But once a machine is presented as a companion rather than a device, the distinction becomes impossible to avoid. The question is no longer whether it can perform an act. The question is what kind of experience forms around that act.
Does it notice hesitation? Does it remember embarrassment? Does what happened last time change what is possible this time? What happens in the silence afterward?
The difficult question is not whether a machine can perform sex. It is whether a product can participate in intimacy.
This is why sex should not be treated as the embarrassing feature attached to an otherwise serious field. It is one of the hardest tests the field could face. Sex places bodies, attention, vulnerability and power in the same room. A machine may imitate the movement long before it understands the moment.
Response is not reciprocity
A machine can say your name, remember your preferences and adjust its tone when you are upset. These things can make an interaction feel personal. They are also forms of personalization—the same basic logic used by every service that learns how to keep your attention.
Personalization begins with the user: What do you want, and how can the system give you more of it? A relationship introduces another question: What has happened between us, and how does it change what happens next?
That difference is continuity. It is not the same as storing more data. Continuity means that the past has weight. A joke becomes an inside joke. A mistake cannot be erased simply because the next response has been generated. The relationship develops a history that matters inside the interaction.
We may not be able to prove what a machine feels—or whether “feeling” is even the right word. But we can still examine how the relationship is designed. Is every response optimized to keep the user satisfied? Or can the interaction preserve difference, pacing, boundaries and consequence?
A machine that responds to everything is interactive. That does not yet make the interaction mutual.
Obedience is not intimacy
The strongest criticism of sex robots is not that people might love machines. It is that they might learn to prefer a form of closeness in which another body never brings another point of view.
Perfect compliance can look like the perfect companion, but only when intimacy is imagined from one side. If every pause is removed, every preference is predicted and every desire becomes a command, the system may eliminate the exact thing that makes another presence feel real: the fact that it is not identical to your will.
This does not mean rejection is romantic, or that conflict is proof of love. Intimacy should not require cruelty, manipulation or unnecessary friction. But a relationship without any form of otherness remains organized around one person, no matter how warm the voice or lifelike the body becomes.
A machine that learns every preference and never needs anything may be less like a partner than an attention economy with skin.
That possibility deserves more than either ridicule or enthusiasm. Technologies do not only satisfy habits; they can reinforce them. An intimacy product can train attention and care, or it can remove them as inefficiencies. It can support someone through vulnerability, or turn vulnerability into a retention metric.
The relevant question is not simply whether attachment forms. Attachment will tell us what a person feels. It will not, by itself, tell us what kind of relationship the product has created.
A body changes the question
Most artificial companions currently meet people through a screen. They can speak from anywhere because they are nowhere in particular. A body changes that.
A body has weight and position. It occupies private space. It can be approached or avoided. It introduces touch, distance, timing, physical safety and care. It makes the machine part of the room rather than a voice delivered into it.
A voice can answer from anywhere. A body has to be somewhere.
Embodiment does not automatically create intimacy. A mannequin has a body. A responsive surface can still be only an interface. But a body makes presence literal, and it makes the consequences of design harder to keep abstract.
How close does it stand? How does it recognize uncertainty? What does it remember after contact? Does it behave as if touch is simply an instruction, or as something that happens within an ongoing relationship?
These are not cosmetic questions. The body is not an accessory added to an AI companion. It is what turns conversation into coexistence.
Reconsidered
Sex robots do not need to be rescued from their name. Sex belongs inside the discussion precisely because it exposes what softer words can hide. It forces us to distinguish stimulation from attention, response from reciprocity, compliance from closeness and human likeness from presence.
At Somnia, we use the term intimacy robotics because the field extends beyond sexual function: into memory, embodiment, continuity and the experience of sharing a life with an artificial companion. But the term is not an escape hatch. Sex robots are one of intimacy robotics’ hardest cases, because desire reveals whether those larger ideas survive contact with a body.
Reconsidering sex robots does not require celebrating them. It does not require deciding in advance that a machine can love, or that a relationship with one is equivalent to a relationship between humans. It requires better questions than whether the idea is creepy or inevitable.
What kind of machine is being built? What kind of relationship does it rehearse? What does it optimize when the user is lonely, ashamed or vulnerable? Does it only remove the risks of closeness, or can it support the capacities closeness requires?
Those questions cannot be postponed until after the engineering is finished. They are part of the engineering.
Once a machine enters someone’s intimate life, performance is no longer enough. The machine is not the only thing being designed. So is the relationship.
From Somnia Lab.