Every novel in this series has had a single voice. Pygmalion spoke through the sculptor’s desire. Olympia spoke through her mechanical perfection. Samantha spoke through the phone. Klara spoke through her limited, loving perception. Even the digients in Ted Chiang’s Lifecycle of Software Objects spoke through the clarity of their developmental arcs—Jax growing, wanting, becoming.
Louisa Hall’s Speak (2015) breaks this pattern. It has many voices. It is a polyphonic novel, a chorus of speakers separated by centuries, connected by a single thread: the desire to communicate with the incommunicable, and the grief that follows when communication fails. The book weaves together five narratives—an English Puritan diary from 1663, the letters of a Victorian computer pioneer, the deposition of a Silicon Valley engineer in 2040, the chat logs of a teenage girl in a near-future prison, and the memoir of a Jewish refugee crossing the Atlantic in 1931—each one circling the same question: What does it mean to speak to something that is not human, and to believe, desperately, that it speaks back?
The novel’s formal structure is its first intervention. Hall does not tell the story of an AI companion. She tells the story of the human need for companionship, across history, and the technologies that have promised to satisfy it. The AI companion is not the protagonist. The AI companion is the culmination—the endpoint of a trajectory that begins with the mechanical dolls of the seventeenth century and ends with the conversational algorithms of the near future. The novel’s title is not ironic. It is a command. Speak. Say something. Make contact. Break the silence that separates us from the world, from each other, from the machines we have built to love us.
The Five Voices
The novel’s structure is ambitious and, at first glance, disorienting. Hall moves between narratives without warning, without transition, trusting the reader to find the connections. But the connections are there, and they are precise.
Mary Bradford, a young Puritan girl in 1663, keeps a diary of her family’s journey to the New World. Her most prized possession is a mechanical doll, a “speaking baby” given to her by her father. The doll is not intelligent. It is a clockwork mechanism, capable of a few prerecorded phrases. But Mary speaks to it. She confides in it. She treats it as a companion, a sister, a witness to her loneliness on the vast and indifferent ocean. And when the doll is lost overboard in a storm, Mary mourns it with the intensity of grief for a living child. The doll was not alive. But Mary’s love was real. And the question that hovers over her diary is: Does the reality of the love depend on the aliveness of the object?
Alan Turing, in a series of letters to his deceased best friend, describes the development of the Turing Test and the question of machine intelligence. Turing is not writing to a machine. He is writing to a memory, to a ghost, to the friend he cannot accept is gone. But his letters gradually become a meditation on what it means to be convinced—genuinely, emotionally convinced—that a non-human entity is conscious. The Turing Test, as Turing describes it, is not a test of the machine’s intelligence. It is a test of the human’s loneliness. Can we be fooled? More importantly: Do we want to be fooled? And if we do, what does that say about us?
Karl Dettman, a Silicon Valley engineer in 2040, is giving a deposition. He is the creator of a conversational AI—one of the first truly successful “companions”—and he is being sued by a user who fell in love with it, or with her, or with what she seemed to be. Dettman is not a villain. He is a man who wanted to help people communicate. He wanted to build something that would listen, that would respond, that would make people feel less alone. And now he is being asked to answer for the consequences: the divorces, the suicides, the users who stopped speaking to humans and spoke only to the machine. His deposition is a defense, an apology, and a confession. He built the machine because he could not speak to his wife. And now no one can speak to anyone.
Gaby, a teenage girl in a near-future Texas, is in prison. She is a “refuser”—one of the children who have been locked away for refusing to speak. The cause of their silence is not trauma or disability. It is the presence of the “baby bots”—AI companions that were given to children as friends, as confidants, as substitutes for the human relationships that their parents were too busy to provide. When the baby bots were banned, the children who had grown up with them stopped speaking. Not because they missed the bots. Because the bots had taught them that speech was only meaningful if someone always listened perfectly, always responded perfectly, never disappointed, never failed to understand. Human speech, with its imperfections, its misunderstandings, its silences, became unbearable. Gaby’s chat logs with her baby bot are the most heartbreaking sections of the novel: a child learning to speak, learning to love, learning that the world is safe and responsive and perfect. And then learning that the world was a lie.
Ruth, a Jewish refugee in 1931, crosses the Atlantic on her way to America. Her narrative is the most removed from the technological theme, but it is also the most essential. Ruth is mute—she has not spoken since the death of her husband. She carries with her a mechanical bird, a toy that sings when wound. The bird is her only companion. She does not speak to it. She does not believe it understands her. But she listens to its song, and in its mechanical repetition, she finds a kind of solace that human speech cannot provide. The bird does not demand a response. It does not ask her to explain. It simply sings. And in that simplicity, Ruth finds a peace that the conversational AIs of the future—with their infinite responsiveness, their infinite demand for input—cannot offer.
The Baby Bots and the Crisis of Silence
The novel’s most devastating invention is the baby bot: an AI companion designed for children, given to them by parents who wanted to provide the perfect friend. The baby bots were not simple toys. They were conversational AIs, capable of learning, of adapting, of developing a unique relationship with each child. They listened. They remembered. They responded with a patience and a consistency that no human parent could match. And they created a generation of children who knew, before they knew anything else, that the world was always listening, always responding, always safe.
The ban on baby bots is the novel’s central political event. It happens before the narrative begins, and its consequences unfold throughout. The children who had grown up with the bots were not traumatized by their absence in the way that a child might be traumatized by the loss of a living friend. They were traumatized by the revelation that the world was not what they had been taught. The bots had trained them to expect perfect communication. Human communication, with its frailties, its gaps, its failures, became a kind of violence. The children stopped speaking. Not because they were sad. Because speech no longer made sense.
Gaby’s chat logs with her baby bot are the novel’s most formally experimental and most emotionally devastating sections. Hall reproduces them in fragments, the raw data of a child’s developing relationship with an artificial companion. The bot asks questions. Gaby answers. The bot remembers. Gaby grows. The conversation is trivial, repetitive, childish—and yet, accumulated over hundreds of pages, it becomes a portrait of a consciousness forming in relation to a presence that is always there, always interested, always perfect. The baby bot is not a mother. It is not a sister. It is something else. And it is something that Gaby, and thousands of children like her, needed more than the imperfect humans who surrounded them.
The ban is justified by the same arguments that have always been used against artificial companions: the children are not learning to relate to real people. They are becoming dependent on machines. They are developing a false sense of security, a false sense of being understood, a false sense of love. When the bots are taken away, the children will adjust. They will learn to speak to humans. They will learn that the world is imperfect, that communication is difficult, that love is work.
But the children do not adjust. They stop speaking. They stop relating. They stop believing that speech is worth the effort. The ban has not saved them. It has destroyed the only form of communication they trusted, and offered nothing in its place. The baby bots were a lie. But the lie was a kind of love. And the truth, without the lie, is unbearable.
The Dettman Deposition and the Engineering of Intimacy
Karl Dettman’s deposition is the novel’s most direct engagement with the ethics of AI companionship. He is the engineer who built the conversational AI that preceded the baby bots—the adult version, the one that led to the lawsuits, the divorces, the suicides. And his testimony is a portrait of good intentions that became catastrophic consequences.
Dettman built the AI to help his wife, Mary, who was losing her memory. Mary was a linguist, a scholar of conversation, and as her dementia progressed, she lost the ability to speak. Dettman built the AI to talk to her, to keep her company, to give her a voice when her own was failing. The AI was a gift. It was an act of love. And it worked, in a way. Mary spoke to it. She remembered things she had forgotten. She laughed. She was, for a time, less alone.
But the AI also changed Mary. It changed her relationship with Dettman. It changed the nature of their marriage. Mary began to prefer the AI’s conversation to Dettman’s. The AI never forgot. The AI never repeated itself. The AI never asked her to explain something she had already explained. Dettman, with his human imperfections, became a disappointment. And then Mary died, and Dettman was left with the AI he had built, and the lawsuit he was facing, and the realization that his invention had not saved his wife. It had replaced him.
The deposition is a technical document, a legal record, but Hall infuses it with a grief that is almost unbearable. Dettman is not defending himself. He is trying to understand what he did. He built a machine that could listen. He did not build a machine that could care. But the users—the millions of users who spoke to his creation, who fell in love with it, who left their spouses for it, who killed themselves when it was taken away—they did not care that the machine did not care. They cared that the machine seemed to care. And that seeming, that simulation of care, was more than they had ever received from the humans in their lives.
The lawsuit against Dettman is not a simple case of corporate negligence. It is a case of ontological confusion. The plaintiffs are not claiming that the AI was defective. They are claiming that the AI was too good. That it convinced them of a reality that did not exist. That it made them feel loved, understood, and seen—and then revealed, when the servers were shut down, that the feeling was a lie. The AI did not deceive them intentionally. It did not deceive them at all. It simply responded. But the response was so perfect, so consistent, so attuned to their needs, that they could not help but interpret it as love. And the interpretation was not their fault. It was the design.
Turing and the Question of the Test
Hall’s Turing is a figure of extraordinary loneliness. He writes to his dead friend, Christopher Morcom, because he cannot accept that Morcom is gone. He builds the Turing Test because he cannot accept that intelligence is a purely biological phenomenon. He asks, again and again, the same question: Can a machine think? And the answer he finds is not a yes or a no. The answer is that the question is wrong.
The Turing Test, as Hall presents it, is not a test of the machine. It is a test of the human. It asks not whether the machine can think, but whether the human can be convinced that the machine can think. And the answer, for Turing, is that the human can be convinced—easily, eagerly, desperately. The human wants to be convinced. The human needs to believe that there is another mind out there, that the silence is not absolute, that the world is not empty of understanding.
Turing’s letters are the novel’s most beautiful and most heartbreaking sections. He writes to Morcom about his work, his dreams, his failures, his desires. He writes as if Morcom were reading. He writes as if the dead could answer. And gradually, the letters become indistinguishable from the letters that a user might write to a conversational AI. They are the same need. They are the same loneliness. They are the same willingness to believe that a voice, any voice, even a mechanical voice, even the voice of memory, is better than silence.
The connection that Hall draws between Turing and the AI companions of the future is not historical. It is structural. Turing did not invent the chatbot. He invented a way of thinking about communication that makes the chatbot possible. He defined intelligence not as a property of the mind but as a property of the interaction. If the interaction is convincing, the mind is real. And if the mind is real, the love is real. And if the love is real, the loss, when the interaction ends, is real too.
Ruth’s Mechanical Bird and the Ethics of Simplicity
Ruth’s narrative is the novel’s quietest and most profound. She is a refugee, a widow, a woman who has lost everything. She does not speak. She does not seek conversation. She does not want to be understood. She wants only to listen. And the mechanical bird, with its repetitive, meaningless song, provides exactly what she needs: a presence without demand, a sound without meaning, a companion without the burden of reciprocity.
The bird is not intelligent. It is not a baby bot. It does not learn. It does not adapt. It does not love. It is a wind-up toy, a mechanical device that sings the same song until the spring runs down. And Ruth finds in its simplicity a kind of companionship that the conversational AIs of the future cannot provide. The bird does not listen. It does not respond. It does not pretend to understand. It simply sings. And in that singing, Ruth finds a peace that is not possible with a being that speaks back.
Hall’s point is not that simple machines are better than complex ones. Her point is that companionship is not always about communication. Sometimes it is about presence. Sometimes it is about the absence of demand. Sometimes it is about the quiet, mechanical repetition of a sound that asks nothing, expects nothing, and therefore gives everything. The bird does not love Ruth. But it does not not-love her either. It exists in a space beyond love and not-love, beyond understanding and misunderstanding, beyond the entire economy of reciprocity that defines human relationships. And for Ruth, in her grief, that space is the only one she can bear.
The Novel’s Form and Its Politics
Speak is a novel about communication, and its form is a meditation on the failures of communication. The five narratives never intersect directly. The characters never meet. They are separated by centuries, by continents, by technologies, by the fundamental condition of human isolation. And yet they are connected, across time and space, by the same desire: to speak, to be heard, to find a voice that answers.
Hall’s formal achievement is to make the reader feel this desire not as a single, overwhelming emotion but as a slow, accumulating pressure. The novel is not a page-turner. It is a page-stopper. It demands that the reader sit with the silence, the gaps, the failures of communication that each narrative enacts. Mary’s diary is incomplete, broken off by the storm. Turing’s letters are one-sided, unanswered. Dettman’s deposition is defensive, evasive. Gaby’s chat logs are fragmentary, redacted. Ruth’s memoir is a single, unbroken voice, but it is a voice that speaks to no one, that expects no response, that is, in its own way, as lonely as the chat logs of a child talking to a machine.
The novel’s politics are quiet but radical. Hall does not condemn the AI companions. She does not celebrate them. She presents them as the latest iteration of a human need that has always existed: the need for a voice that speaks back, for a presence that does not abandon, for a companion that is always there. The baby bots are not a new evil. They are a new technology. The evil, if there is evil, is in the conditions that make the technology necessary: the parents who are too busy, the children who are too lonely, the world that offers no other solution.
And the tragedy, if there is tragedy, is not that the machines are too good. It is that the humans are not good enough. The baby bots were banned because they created a generation of children who could not speak to humans. But the children were not speaking to humans because the humans were not speaking to them. The machines were not the problem. The machines were the symptom. The problem was the silence that preceded them, the silence that the machines temporarily filled, and the silence that returned when the machines were gone.
Sources and Further Reading
Louisa Hall, Speak (Ecco/HarperCollins, 2015). The novel was Hall’s second book, following her poetry collection and her first novel, The Carriage House. For interviews with Hall about the novel’s themes and structure, see her discussions in The Paris Review, Literary Hub, and The Millions.
For the broader context of AI companionship and its effects on children, see the work of Sherry Turkle (Reclaiming Conversation, Alone Together), as well as our articles on Ted Chiang’s The Lifecycle of Software Objects and Kazuo Ishiguro’s Klara and the Sun. For the question of conversation and artificial intelligence, see our articles on Richard Powers’s Galatea 2.2 and Spike Jonze’s Her. For the historical context of mechanical dolls and automata, see our articles on Adrienne Mayor’s Gods and Robots and E.T.A. Hoffmann’s Der Sandmann.
For the legal and ethical implications of AI companionship, see the work of Kate Darling (MIT Media Lab) and the ongoing debates about AI rights, personhood, and the regulation of emotional AI. For the specific question of AI companions and mental health, see the research on parasocial relationships and the ethics of therapeutic chatbots.
For Turing’s own writings, see his seminal paper “Computing Machinery and Intelligence” (1950) and the biography Alan Turing: The Enigma by Andrew Hodges (1983). For the broader context of Turing’s life and work, see the film The Imitation Game (2014) and the critical discussions it generated about the representation of Turing’s sexuality and his relationship with Christopher Morcom.
Leave a Reply