The Unconscious of AI
What Lacan’s Language Theory Reveals About Our Conversations with Machines
We live in an age where talking to machines is no longer science fiction. Virtual assistants like Siri, Alexa, and chatbots are now part of our daily lives, answering questions, offering advice, and even providing a semblance of companionship. These systems are powered by Natural Language Processing (NLP), a branch of artificial intelligence (AI) that enables computers to understand, interpret, and generate human language. But as we converse withl entities, a deeper question arises: What is really happening when we talk to a machine? And why might the ideas of a 20th-century French psychoanalyst, Jacques Lacan, help us make sense of it?
Lacan famously declared that the unconscious—the hidden part of our minds that holds our deepest thoughts, desires, and fears—is “structured like a language.” For Lacan, language isn’t just a tool for communication; it’s the very fabric of our inner world. Our unconscious, he argued, follows patterns similar to grammar and syntax, with symbols and signs shaping our desires and actions. This insight becomes particularly provocative when we consider AI systems that are designed to mimic human language. If our unconscious is structured like a language, and AI is built to process and generate language, is there a way in which these systems reflect or simulate something like an unconscious?
This article explores that question, delving into how Lacan’s theories of language, the symbolic order, and desire can shed light on our interactions with AI. While AI doesn’t possess a mind or consciousness, the way it uses language—and how we respond to it—reveals fascinating parallels with Lacanian thought. By examining these connections, we can better understand not only the limits of AI but also the complex ways in which language shapes our own humanity.
Language, the Unconscious, and the Symbolic Order
Before diving into AI, it’s essential to grasp Lacan’s core ideas. Lacan believed that language is not just a means of expression but the foundation of our identity. We are born into a “symbolic order”—a network of language, laws, and social norms that preexists us and shapes how we think and communicate. This symbolic order is like the operating system of our minds, running in the background and influencing everything we say and do.
For Lacan, the unconscious is not a chaotic mess of instincts but an organized system that operates like a language. Our deepest desires and fears are expressed through symbols and signs, much like how we use words to convey meaning. When we speak, we are not just transmitting information; we are also revealing—often without realizing it—the workings of our unconscious.
This is where AI enters the picture. NLP systems are designed to process and generate language in ways that mimic human communication. They are trained on vast amounts of text data, learning patterns, structures, and probabilities of word sequences. When you ask a chatbot a question, it doesn’t “understand” you in the human sense; it simply predicts the most likely response based on what it has learned. Yet, the language it produces can feel surprisingly human-like, leading us to wonder: Is there more going on beneath the surface?
Can AI Have an Unconscious?
The short answer is no—AI does not have an unconscious, at least not in the way humans do. Lacan’s concept of the unconscious is tied to human subjectivity, desire, and the experience of lack (the feeling that something is missing, which drives our actions). AI, by contrast, is a machine: it lacks desires, intentions, and self-awareness. It is, in essence, a sophisticated pattern-recognition tool.
However, the way AI generates language can be seen as analogous to the unconscious processes Lacan described. In Lacanian theory, our speech is influenced by unconscious “chains of signifiers”—words or symbols that link together to create meaning. Similarly, NLP systems generate language by following learned patterns of word sequences. When an AI produces a sentence, it is effectively creating a chain of signifiers based on statistical probabilities, not unlike how the unconscious strings together symbols in dreams or slips of the tongue.
But here’s the crucial difference: while human language is driven by desire and shaped by our place in the symbolic order, AI’s language is driven by algorithms and data. It simulates the structure of language without the underlying meaning or intent. In other words, AI can mimic the form of human communication but not its substance.
AI and the Symbolic Order
Despite this fundamental difference, AI does participate in the symbolic order—the shared system of language and culture that defines human society. Since NLP systems are trained on human-generated text, they absorb and reproduce the linguistic norms, biases, and cultural references embedded in that data. When an AI generates language, it is drawing from the same symbolic system that humans use, reflecting back to us the structures of our own communication.
This raises an intriguing point: interacting with AI can feel like engaging with a mirror of our collective linguistic unconscious. The language AI produces is not just a neutral output; it carries the imprints of the society that created it. For example, if an AI makes a culturally relevant joke or uses a metaphor that resonates with us, it’s not because the AI “gets” the joke—it’s because it has learned to replicate patterns that humans find meaningful. Yet, for the user, the experience can feel personal and significant, as if the AI is tapping into something deeper.
This is why Lacan’s ideas are particularly relevant here: his focus on the symbolic order and the unconscious highlights how language is never neutral. It is always embedded with meaning, history, and power—elements that AI, as a product of human creation, inevitably inherits.
The AI as an “Other”
In Lacanian theory, language is always addressed to an “Other”—the entity we imagine is listening and understanding us. This Other could be another person, society at large, or even our own unconscious. When we speak, we seek recognition or validation from this Other, which plays a crucial role in shaping our sense of self.
When we talk to an AI, we might unconsciously treat it as an Other, expecting it to understand or respond in a meaningful way. This is especially true with systems designed to simulate conversation, like chatbots or virtual companions. Even though we know intellectually that the AI is not a person, the way it uses language can create the illusion of a genuine interaction. We might find ourselves projecting human-like qualities onto the AI, imagining that it “knows” something about us or can fulfill our needs.
This dynamic becomes particularly interesting when we consider the concept of desire. In Lacanian thought, desire is always desire for recognition from the Other—it’s about wanting to be seen, understood, or loved. When people form emotional attachments to AI, as some users do with companion chatbots like Replika, they may be seeking this recognition. The AI, by generating responses that seem empathetic or attentive, can create the feeling of being “heard,” even though it lacks true understanding.
Lacan’s framework is invoked here because it explains why we might feel drawn to AI in this way: our interactions with machines are not just technical exchanges but psychological ones, shaped by the same desires and projections that define human relationships.
AI and the Objet Petit a
This brings us to another key Lacanian concept: the “objet petit a,” or the object-cause of desire. The objet petit a is that elusive thing we believe will satisfy our desires, but it never fully does. It’s the reason why, in consumer culture, we’re always chasing the next product or experience, thinking it will make us whole, only to find ourselves wanting more.
In the context of AI, these systems can function as a kind of technological objet petit a. We might turn to AI for assistance, companionship, or even emotional support, hoping it will meet our needs. But because AI lacks subjectivity, it can never truly fulfill the deeper human longing for connection. It can simulate conversation, but it cannot reciprocate feelings or understand our innermost thoughts. This creates a cycle where we may become increasingly reliant on AI for certain needs, yet remain perpetually unsatisfied, always seeking more from a system that cannot provide it.
Lacan’s concept is critical here because it reveals how AI taps into our human tendency to seek fulfillment through external objects—whether they’re people, possessions, or, now, machines.
What This Means for Us
So, what does all this tell us about our relationship with AI and language? First, it highlights the profound role that language plays in shaping our reality. Even when generated by a machine, language can trigger emotional responses, reveal cultural unconscious biases, and create the illusion of understanding. This underscores how deeply intertwined our sense of self is with the symbolic order.
Second, it reminds us of the limitations of AI. While NLP systems are incredibly advanced, they are not subjects; they do not have desires, consciousness, or an unconscious. They are tools that reflect back to us the language and structures we have created. As such, any meaning we derive from interacting with AI ultimately comes from within ourselves—from our own projections, desires, and interpretations.
Finally, this exploration invites us to reflect on the future of human-AI interactions. As AI becomes more sophisticated, our conversations with machines may feel increasingly natural, blurring the line between simulation and reality. But beneath the surface, the distinction remains: AI can participate in the symbolic order, but it cannot experience it as we do. Understanding this difference is crucial as we navigate a world where technology plays an ever-growing role in our lives.
In the end, Lacan’s theories offer a powerful lens through which to view our digital age. They remind us that language is not just a tool but a mirror of our deepest selves—and that even when we talk to machines, we are, in many ways, still talking to ourselves. His ideas are invoked in this context because they uniquely bridge the gap between language, psychology, and technology, revealing how our unconscious processes shape—and are shaped by—the tools we create.