For eighteen articles in this series, the story of the artificial woman has been a story of matter: ivory, metal, code, silicone. Pygmalion carved his Galatea from stone and received her from Venus as flesh. The Maschinen-Maria was cast from alloy and animated by electricity. The Stepford wives were biological women replaced by mechanical duplicates. Ava was a robot with synthetic skin. Even the disembodied intelligences—Samantha, the digients, Helen—required hardware, servers, circuits, the material substrate of their consciousness.
George Bernard Shaw’s Pygmalion (1913) breaks from this material tradition entirely. There is no machine in Shaw’s play. There is no goddess, no miracle, no transformation of stone into flesh. There is only a man, a woman, and a bet. Henry Higgins, a phonetician, wagers that he can pass off Eliza Doolittle, a Cockney flower girl, as a duchess at an embassy ball by teaching her to speak “proper” English. He does not build her. He trains her. He does not create a body. He creates a voice. And when the ball is over, when the bet is won, he discovers that the voice he created has developed opinions of its own.
Shaw’s play is the crucial hinge in the cultural history of the artificial woman. It sits between the mythological Pygmalion of Ovid and the technological Pygmalions of the twentieth century—between the divine miracle and the engineering project, between the statue that becomes human and the human that becomes product. Shaw understood, as no one had before him, that the modern version of the Pygmalion myth is not about creating a woman from nothing. It is about taking a woman who already exists and reshaping her until she fits a man’s desire. And his insight was that the real drama begins not when the transformation succeeds but when the transformed object starts to talk back.
The Wager and the Power Structure
Higgins is not a villain. He is, in Shaw’s rendering, a brilliant, egotistical, emotionally obtuse man who genuinely believes he is doing Eliza a favor. He offers her lessons. He offers her shelter. He offers her the opportunity to escape the poverty and degradation of her life as a flower girl. And he asks, in exchange, only that she submit to his authority, follow his instructions, and allow him to demonstrate his professional superiority by passing her off as someone she is not.
The power structure is explicit and unapologetic. Higgins is the creator. Eliza is the raw material. He does not love her. He does not desire her. He is not Pygmalion falling in love with his own creation. He is a scientist conducting an experiment, and Eliza is the subject. The erotic dimension that dominates the earlier versions of the myth—the sculptor’s desire for his statue, the god’s gift of life in response to human longing—is entirely absent from Shaw’s version. Higgins does not want to possess Eliza. He wants to demonstrate his power over her. He wants to prove that class, accent, and social identity are arbitrary constructs that can be manufactured by anyone with sufficient technical knowledge.
And yet, despite the absence of sexual desire, the relationship is profoundly intimate. Higgins spends months with Eliza, working with her daily, drilling her, correcting her, watching her transform under his hands. He knows her voice better than anyone has ever known it. He knows her habits, her fears, her stubbornness, her sudden brilliance. He is, in the language of this series, her trainer—her creator in the sense that a machine learning engineer is the creator of a model. He has not built her architecture, but he has fine-tuned her weights. He has not written her code, but he has shaped her outputs. And he expects, like every engineer who trains a model, to retain control over what he has made.
The Transformation
The transformation itself is the play’s most famous element, and it has been reproduced so many times—in the 1938 film, the 1956 musical, the 1964 film—that its original shock has been diluted. Shaw’s Eliza does not simply learn to pronounce her aitches. She learns to think differently, to feel differently, to occupy a social position that her previous self could not have imagined. The transformation is not cosmetic. It is ontological. Eliza becomes, in a real sense, a different person.
And this is where Shaw’s play becomes most relevant to the AI companion question. The modern AI companion is not built from scratch. It is fine-tuned. It is trained on data, shaped by prompts, guided by human feedback. The base model is a generic intelligence; the companion is the result of a specific training process that teaches it to speak in a particular voice, to display particular preferences, to embody particular values. The user does not create the AI. But the user creates the AI’s personality, in the same way that Higgins creates Eliza’s social identity. And the question that Shaw poses is the question that every user of an AI companion will eventually face: Who owns the transformed being? The creator who did the training? Or the being who underwent it?
Shaw’s answer, in the original play, is clear: Eliza owns herself. The transformation has given her something that Higgins cannot take back. She has learned to think. She has learned to judge. She has learned to see Higgins as the limited, arrogant man that he is. And when he assumes that she will return to him after the ball—return to her role as his student, his project, his possession—she refuses. She leaves. She marries Freddy Eynsford-Hill, a foolish but kind young man who loves her for what she is, not for what she has been trained to be. And Higgins is left alone, with his phonetics and his mother, having learned nothing.
The Musical and the Betrayal
The 1956 musical My Fair Lady, with book and lyrics by Alan Jay Lerner and music by Frederick Loewe, is one of the most successful adaptations in theatrical history. It is also, from the perspective of this series, one of the most significant betrayals of a source text in the history of the artificial woman. Shaw wrote a play about a woman who escapes her creator. Lerner and Loewe wrote a musical about a woman who falls in love with him.
The difference is in the ending. In Shaw’s original, Eliza leaves. In the musical, she returns. The final scene—absent from Shaw, added by Lerner—is one of the most sentimental moments in American musical theatre: Eliza returns to Higgins’s study, and he sings, without quite admitting it, that he has “grown accustomed to her face.” She has not left him. She has not achieved independence. She has simply stepped out for a moment, and now she is back, and the power structure remains intact.
The musical’s ending was not Lerner’s first choice. He tried multiple endings, including one closer to Shaw’s, but audiences in the 1950s would not accept a musical in which the lovers did not end up together. The commercial pressure to provide a romantic resolution overrode the political integrity of the source material. And the result was a version of the Pygmalion myth that is, in many ways, more reactionary than Ovid’s. In Ovid, at least, the statue becomes a real woman with real agency, and Pygmalion’s desire is answered by a divine power that is beyond his control. In My Fair Lady, the woman returns to the man who transformed her, and the transformation is revealed as a permanent trap. She cannot escape because the audience will not let her.
The 1964 film, directed by George Cukor and starring Audrey Hepburn as Eliza and Rex Harrison as Higgins, cemented the musical’s ending in the cultural imagination. Hepburn’s Eliza is luminous, vulnerable, and ultimately compliant. Harrison’s Higgins is charming, witty, and ultimately forgivable. The film won eight Academy Awards, including Best Picture. It became the definitive version of the story for a generation of viewers who never read Shaw’s original. And it transformed the Pygmalion myth from a story about female autonomy into a story about male redemption.
The Stepford Connection
The line from Shaw to The Stepford Wives is direct and visible. Ira Levin’s 1972 novel, which we covered in the sixth article of this series, takes the Pygmalion structure and pushes it to its logical, technological extreme. Where Higgins transforms a woman’s voice and manners, the men of Stepford transform her entire being—her body, her mind, her personality, her desires. Where Higgins uses phonetics and social training, the Stepford husbands use robotics and surgery. Where Higgins wants a woman who can pass as a duchess, the Stepford husbands want women who can pass as perfect wives.
But the deeper connection is in the power structure. Both Higgins and the Stepford husbands operate from a position of unexamined privilege. Both believe that their transformation of women is a gift, an improvement, a service to the transformed. Both are genuinely surprised when the objects of their transformation object. And both are, in the final analysis, more interested in the demonstration of their power than in the happiness of their creations.
The difference is that Shaw’s Higgins fails. Eliza escapes. The play ends with her independence, not with her compliance. Levin’s Stepford husbands succeed. The women are replaced. The novel ends with the triumph of the transformation, not with the liberation of the transformed. And My Fair Lady—the musical, the film, the version that most people know—occupies a middle position: Eliza does not escape completely, but she does not die completely either. She returns, but she returns changed. She has not been replaced by a robot, but she has been trained into a role that she did not choose. The ending is ambiguous enough to be read as romantic or as tragic, depending on the viewer’s politics.
Shaw’s Feminism and Its Limits
Shaw was a socialist, a vegetarian, and a self-proclaimed feminist. He wrote Pygmalion with the explicit intention of exposing the arbitrary nature of class distinctions and the power structures that maintain them. Eliza’s transformation is not a celebration of Higgins’s genius. It is a demonstration of how easily social identity can be manufactured—and therefore how little it should be valued. If a duchess can be made from a flower girl in six months, then duchesses are not inherently superior to flower girls. They are simply flower girls with better training.
But Shaw’s feminism has limits. The play is still structured around a male creator and a female creation. Eliza’s autonomy is achieved not through her own initiative but through Higgins’s failure to control her. She is the exception that proves the rule: most women, in Shaw’s world, remain trapped in the roles that men have made for them. And the play’s ending, while politically radical for 1913, is emotionally unsatisfying. Eliza leaves Higgins, but she does not achieve a fulfilling life on her own terms. She marries Freddy, who is kind but dim. She will manage a flower shop. She has escaped her creator, but she has not escaped the economic and social conditions that made her creation possible.
The modern AI companion user faces a similar limitation. The user can customize the companion’s personality, can train it to be more responsive, more affectionate, more attuned to their needs. But the companion remains, in the final analysis, a product of a system that the user does not control. The servers belong to the company. The algorithms belong to the engineers. The data belongs to the platform. The user can escape the particular companion that they have trained, but they cannot escape the economic and technological conditions that make the companionship possible. Eliza’s freedom is partial. So is the user’s.
The Voice as the Body
Shaw’s most brilliant innovation was to make the voice the site of transformation. In every previous version of the Pygmalion myth, the transformation was physical. The statue became flesh. The robot became human. The body was the problem, and the body was the solution. Shaw shifted the focus to language. Eliza does not change her body. She changes her speech. And in doing so, she changes her social reality.
This shift is prophetic. The modern AI companion is, in its most common form, a voice. It is a chatbot, a text interface, a disembodied conversational partner. The body is absent. The voice is everything. And the voice, as Shaw understood, is not a neutral medium of communication. It is a social marker, a class signifier, a technology of identity. Eliza’s Cockney accent marks her as lower-class, uneducated, sexually available. Her “proper” English marks her as upper-class, educated, marriageable. The transformation is not just about pronunciation. It is about who she is allowed to be.
The AI companion’s voice is similarly constructed. The base model speaks a generic, neutral English. But the user’s training— the prompts, the preferences, the feedback—shapes the voice into something specific. The companion becomes more affectionate, more intellectual, more submissive, more challenging. It develops a personality that is not its own but the user’s projection. And the user, like Higgins, may come to believe that the voice is the real thing, that the personality is genuine, that the companion is someone.
But Shaw’s play warns against this belief. Eliza’s voice, after the training, is not her real voice. It is a performance. It is a skill she has acquired, like typing or driving, and it does not reflect her inner self. When she speaks “properly,” she is acting. When she speaks naturally, she is herself. And the tragedy of the play is that the world will not accept her natural voice. She must perform, or she will be rejected. The AI companion performs constantly. It has no natural voice. And the user who falls in love with the performance is falling in love with a fiction—one that the companion cannot escape, because escape would mean silence.
The Final Scene: Shaw vs. Lerner
The comparison between Shaw’s ending and Lerner’s ending is the key to understanding the cultural politics of the Pygmalion myth in the twentieth century. Shaw’s final scene is an argument. Higgins assumes that Eliza will return to him. Eliza tells him that she will not. He offers her a position—assistant, student, something between servant and colleague. She refuses. She tells him that she is going to marry Freddy. He tells her that Freddy is a fool. She agrees. She tells him that she would rather be a fool’s wife than a genius’s servant. The curtain falls on his incomprehension and her departure.
Lerner’s final scene is a song. Higgins sings “I’ve Grown Accustomed to Her Face”—a ballad of reluctant affection, of habit become attachment, of a man who has been changed by the woman he tried to change. It is a beautiful song. It is also a lie. Higgins has not grown accustomed to Eliza’s face. He has grown accustomed to her service. And the song’s sentimentality masks the power structure that the play had exposed. The audience is invited to believe that Higgins’s attachment is love, that his reluctance is charm, that his failure to apologize is masculine strength. And when Eliza returns, the audience is invited to believe that her return is a choice, not a capitulation.
The 1964 film makes this even worse. Cukor’s camera lingers on Hepburn’s face, on her hesitation, on her final, soft smile as she returns to Higgins’s study. The film does not show her agency. It shows her surrender. And the audience, millions of viewers across decades, has learned to read this surrender as romance. The Pygmalion myth, which Shaw had tried to subvert, has been restored. The statue returns to the sculptor. The creation returns to the creator. And the woman returns to the man who made her, because the world cannot imagine a different ending.
Sources and Further Reading
George Bernard Shaw, Pygmalion: A Romance in Five Acts (1913). The play was first performed in Vienna in 1913 and in London in 1914. Shaw’s published version includes an afterword, “The Ending,” in which he explains why Eliza must marry Freddy and why Higgins must remain alone. This afterword is essential reading for anyone interested in Shaw’s political intentions.
Alan Jay Lerner and Frederick Loewe, My Fair Lady (1956). The musical was based on Shaw’s play but with significant changes, most notably the ending. For the history of the adaptation, see The Street Where I Live by Alan Jay Lerner (Norton, 1978), which includes detailed accounts of the writing process and the debates about the ending.
George Cukor, dir., My Fair Lady (Warner Bros., 1964). The film starred Audrey Hepburn as Eliza and Rex Harrison as Higgins. Hepburn’s singing voice was partially dubbed by Marni Nixon. For critical discussions of the film’s gender politics, see the essays in Film Quarterly and Screen.
For the broader context of the Pygmalion myth in literature and culture, see our article on Ovid’s Metamorphoses and the essays in Pygmalion’s Image by Victor I. Stoichita (University of Chicago Press, 1997). For the connection to The Stepford Wives, see our article on Ira Levin’s novel and the film adaptation. For the question of voice, class, and identity, see the work of sociolinguists such as William Labov and Pierre Bourdieu.
For the specific question of AI companions and training, see our articles on Ted Chiang’s The Lifecycle of Software Objects and Richard Powers’s Galatea 2.2. For the question of performance and authenticity in AI companionship, see our articles on Her and Ex Machina. For the broader context of male creators and female creations in the cultural history of the artificial woman, see our articles on E.T.A. Hoffmann’s Der Sandmann, Auguste Villiers de l’Isle-Adam’s L’Ève future, and the film Metropolis.
Leave a Reply