The ELIZA Archaeology Project is putting the first chatbot back under the microscope, with an interdisciplinary group of scholars, artists and programmers studying Joseph Weizenbaum’s 1960s MIT program and the ideas it kicked loose.
The project describes ELIZA as the original and highly influential text-based chatbot, created by Weizenbaum at MIT as part of Project MAC. That history matters because ELIZA helped establish a style of human-computer interaction that is now everywhere: people typing to software agents and expecting something that resembles conversation to come back.
The group is not treating ELIZA as a museum trinket. It has published an accurate reimplementation of ELIZA so readers can try the program rather than absorb another round of chatbot folklore. That is the correct instinct. With ELIZA, the gap between what the program does and what people think it does has always been part of the story.
What the project is studying
According to the project, its work will cover ELIZA’s history, context and code. The team says it plans to explain how the program works, examine the programming culture around Weizenbaum, and trace his later shift from building ELIZA/DOCTOR to warning about the risks of treating machines as if they were human.
That last point is not incidental. ELIZA’s influence is not only technical. The project says the program shaped how people began to imagine computers as systems that could take part in conversation. That idea now sits beneath much of the public argument over AI, even when the implementation has changed beyond recognition.
The project also plans to study later works influenced by ELIZA and to consider how talking computer programs have been represented in literature and film. In other words, the archive is not limited to code listings and terminal sessions. It also includes the cultural residue left by a program that made people argue with a machine and then argue about what that meant.
A collaborative archive and a book
The ELIZA Archaeology Project says it is being built by investigators with different interests and voices. The team is adding material to the project website as it works through the materials, following collaborative models such as the 2015 Reading Project at Iowa.
The group is also working on a collectively authored book titled Inventing ELIZA: How the First Chatbot Shaped the Future of AI. The project maintains a blog for current findings and points readers to Reading ELIZA: Critical Code Studies in Action for more background.
The pitch is modest by present-day AI standards, which is a relief. The project is not promising a new assistant, a benchmark trophy or a venture-funded oracle. It is trying to read an old program carefully, in context, before yet another generation pretends chatbots began yesterday.
This story draws on original reporting from Daring Fireball.
