Miles Lee

PARRY, created in 1972 by psychiatrist Kenneth Colby, was a chatbot created to simulate a paranoid schizophrenic person. The script that PARRY ran was based on the image of someone suffering from schizophrenia and used a crude and basic understanding of the behaviors of someone with said condition as well as generic judgments and stereotypes, and was built to study the interactions between people and those suffering from a severe mental illness. PARRY was seen as a counterpart to an earlier chatbot, ELIZA – while ELIZA was a simulation of a somewhat empathetic psychotherapist, PARRY was the patient. While both were meant to simulate conversations between the chatbot and the user in a human manner, PARRY’s chatting system was more complex. While ELIZA relied on very basic pattern-matching and would often simply mirror user input, PARRY used a more sophisticated model that had assumptions, intense emotions, and triggers similar to those with schizophrenia. PARRY could also express complex emotions towards users, such as suspicion or fear.

Outside of its technical advantages over other early chatbots, PARRY presents as an interesting use case for chatbots; while ELIZA has a comforting effect that many crave, PARRY often actively makes users uncomfortable and afraid. The program tapped into uncharted territory as far as what we can simulate, and more importantly, why we choose to simulate certain things. While PARRY can in theory help us understand and empathize with mental health patients, it can be criticized for its rudimentary understanding of paranoid schizophrenia and the fearful nature of the chatbot. PARRY stands as a unique example of the capabilities of chatbots and those who create them.

This Case Study is part of a larger essay, Falsities and Detachment: The Rise of AI Chatbots and the Commodification of Human Connection.

---