ELIZA is a retro chatbot, based on the original ‘ELIZA’ script that simulates the communication style of a psychotherapist. It was created at MIT by Joseph Weizenbaum in the 1960s.
ELIZA – A Computer Program For the Study of Natural Language Communication Between Man and Machine
Joseph Weizenbaum, 1966
Table of Contents
Chat with ELIZA
ELIZA WordPress Widget
I wrote the above ELIZA widget as a WordPress plugin that utilizes PyScript, MicroPython, and WebAssembly. It operates in the browser, doesn’t use cookies, and doesn’t send any data to the server, ensuring complete privacy. Thanks to MicroPython, it’s also lightweight and loads quite quickly.
ELIZA in Microsoft Store
After I wrote the ELIZA widget for WordPress, I noticed that there was no classic ELIZA chatbot in the Microsoft Store. I’ve fixed that and made it available there as well:
https://www.microsoft.com/store/productId/9NQ4V4HQB4N2?ocid=pdpshare
Re-implementing old chatbots is kind of a time travel; who knows, maybe one day I’ll re-implement PARRY from 1972, the paranoid schizophrenic companion to ELIZA, laying the foundations for a chatbot museum.
ELIZA in Google Play Store
In March 2024, I released ELIZA 1966 – The First Chatbot for Android as well:
https://play.google.com/store/apps/details?id=com.szymonjessa.eliza
ELIZA is not only the first chatbot, but it’s also the first application I’ve published on the Google Play Store.
The ELIZA effect
Since 1966, ELIZA not only tells a story about computers or AI, but even more — it reveals something crucial about us, humans. Invite a friend to sit down with ELIZA or ChatGPT and observe the conversation. Does your friend attribute human traits to the chatbot? Do they formulate answers as if the computer would think or feel? Or as if it were actually a person that understands sentences like another human? Try to figure it out.
In computer science, the ELIZA effect is the tendency to project human traits — such as experience, semantic comprehension or empathy — into computer programs that have a textual interface. The effect is a category mistake that arises when the program’s symbolic computations are described through terms such as “think”, “know” or “understand.”
Source: Wikipedia – ELIZA effect
No worries, that’s how it is. However, reflect on this in the context of using a computer program for self-reflection, similar to using Virtual Reality. Even if there are glitches and the graphics are poor, our brain fills the gaps, and YOUR emotions and thoughts are still REAL.
ELIZA vs ChatGPT
ELIZA showcases the genius of its authors in the ratio of how much can be achieved by so little code. Reflecting on that small program and its impact, how does the comparison with modern data and compute giant “chatbots” look like?
Name | ELIZA | ChatGPT |
Born | 1966, USA | 2022, USA |
Age | 56 years | 2 years |
Weight | few kB’s 2 000 words in the script | 800GB 175b parameters in the model 45TB training data |
Parents | Joseph Weizenbaum | OpenAI |
Godparents | MIT, DARPA | Angels & Investors |
Family | Human Machine Communication | Large Language Models |
Belief system | Expert & Rules (Word Patterns) | Data & Probability (Word Statistics) |
Beyond these ephemeral reasons, how else can such an old program be used?
NLP Workshop with ELIZA
For the past few years, I have been using ELIZA to conduct NLP & Python workshops, most often at the University of Gdańsk. Students learn about various NLP topics and terminology related to conversational systems, which are still relevant in modern AI solutions and fundamental to the development of chatbots. These topics include NLU, ASR, intents, segmentation, tokenization, pattern matching (with Regular Expressions), natural language generation, the ELIZA effect, and all the essential “tricks” for building conversational systems. Depending on the audience, coding their own ELIZA version in Python is also a possibility. If you find this interesting and believe such a workshop would be beneficial for your needs, please reach out to me. Besides these applications, ELIZA is also excellent for inspiring reflections on AI and human nature in general
Resources
- Jessa, Sz. ELIZA Chatbot Plugin for WordPress. [Software]. https://wordpress.org/plugins/eliza-chatbot/
- Jessa, Sz. ELIZA 1966 – The First Chatbot. [Software].
https://www.microsoft.com/store/productId/9NQ4V4HQB4N2 - Jessa, Sz. ELIZA 1966 – The First Chatbot. [Software]. https://play.google.com/store/apps/details?id=com.szymonjessa.eliza
- Jessa, Sz. ELIZA for MicroPython. [Software]. GitHub. https://github.com/micropython-eliza
- Jessa, Sz. ELIZA for MicroPython. [Software]. PyPI. https://pypi.org/project/micropython-eliza
- MicroPython: https://micropython.org/
- PyScript: https://pyscript.net/
- WebAssembly: https://webassembly.org/
- Weizenbaum, J. (1966). “ELIZA – A Computer Program For the Study of Natural Language Communication Between Man and Machine.” DOI: https://doi.org/10.1145/365153.365168. Full text: https://web.stanford.edu/class/cs124/p36-weizenabaum.pdf
- Weizenbaum, J. (1976). “Computer Power and Human Reason: From Judgment to Calculation.” DOI: https://dl.acm.org/doi/10.5555/540249.
- Wikipedia. (n.d.). ELIZA effect. Retrieved from https://en.wikipedia.org/wiki/ELIZA_effect