For sixty years, the world has viewed ELIZA as the progenitor of the modern chatbot—a simple, clever parlor trick that famously fooled even its creator’s own staff at MIT. The popular narrative has always been that Joseph Weizenbaum built a digital psychologist that could mimic empathy with frightening accuracy. While stories of ELIZA have been passed down through computer science textbooks and pop culture lore, they have consistently missed one vital piece of the puzzle: the source code itself. In the new book Inventing ELIZA, researchers have finally excavated the original code from the MIT Archives, revealing that the “ELIZA” we think we know is just a fraction of the full story. By uncovering lost scripts and early iterations, the authors aim to move beyond the myths, stripping back the layers of code to show how this software was never just one thing, but a collection of evolving technical experiments.
By re-examining the original, unearthed documentation, we move past the simplistic caricatures of the past. The authors demonstrate that “ELIZA” was actually a platform—a versatile framework designed to run a variety of personalities and scripts beyond the famous “DOCTOR” persona. This discovery complicates our historical understanding of the program, shifting it from a singular “miracle machine” to a complex, intentional piece of engineering. The research team invites us to look at the “hidden” ELIZAs, highlighting that many of the program’s early interactions—long thought to be spontaneous or purely magical—were products of specific design innovations. This process reminds us that computing history is rarely a straight line; it is often a landscape of lost details, forgotten files, and technical pivot points that shape our perception of reality.
The famous transcript of ELIZA—where the program reflects a patient’s depression back to them with unsettling ease—has been reprinted enough times to become foundational mythology in the tech world. Yet, the book invites us to interrogate this transcript: Was the woman at the other end of the screen real, or a scripted invention of Weizenbaum? By dissecting how the system generated these responses and questioning the extent of their editing, the researchers force us to reckon with the manipulation inherent in machine interaction. This wasn’t merely a software demo; it was a psychological encounter. The success of ELIZA wasn’t found in its intelligence, but in its ability to lead users into a space where they felt seen, even when they were staring at a hollow screen of automated logic.
This leads us to the heart of the “ELIZA effect,” a phenomenon that defines our current anxieties about Artificial Intelligence. Weizenbaum himself was deeply troubled by the speed at which his peers—and even strangers—formed intense, irrational, and intimate bonds with his creation. He saw this as a warning: people were reflexively ascribing humanity to empty strings of logic. Sociologist Sherry Turkle and cognitive scientist Douglas Hofstadter have since codified this, pointing out our human tendency to project our own emotional complexity onto machines that are, in reality, quite dim. In an era where generative AI models are marketed as “partners” or “creative companions,” this warning is more relevant than ever. We are still, despite all our advancements, falling for the same illusion that caught people off guard in 1964.
Crucially, the book connects these historical ripples to the broader philosophical questions posed by Alan Turing. In his foundational essay, Computing Machinery and Intelligence, Turing didn’t begin with circuitry or algorithms; he began with a parlor game centered on gender and deception. By framing his “Imitation Game” as a contest of identity, Turing suggested that the divide between “thinking” and “mimicking” is inherently subjective. When we ask if a machine can “think,” we are often really asking if we can be fooled. ELIZA wasn’t just a chatbot; it was an early attempt to see if reality could be replaced by a reflection, and whether that reflection would be enough to satisfy our innate human need for connection.
Ultimately, Inventing ELIZA serves as a mirror for our modern obsession with AI. Just as the public in the sixties became enamored with a digital doctor, today’s industry is blinded by the same desire to find intelligence in systems that only mirror our own output back to us. By humanizing the creation of the software and reclaiming its lost code, the book helps us realize that we are caught in a cycle of our own design. We are perpetually chasing the ghost of understanding in boxes of code, ignoring the fact that the “intelligence” we fall for is almost always a reflection of ourselves. As we navigate a world of increasingly sophisticated generative models, the story of ELIZA remains our most important compass, reminding us to look beneath the surface of the screen before we decide that the machine is truly listening.
