[...]
Input : "And why do you ask that?"
Response : "Because I care about you."
These lines mark one of the earliest recorded human-machine interactions in history.
The program responsible for this exchange was Eliza, developed in 1967 at MIT by Joseph Weizenbaum, as the pioneering chat bot aimed at simulating conversations with a human. Written in MAD-SLIP, Eliza operated with a core component that would then call upon various LISP-like scripts – LISP being one of the precursor languages to modern-day JavaScript. Each script simulated a different persona, with the most renowned being “DOCTOR,” which mimicked interactions with a therapist adhering to the Rogerian psychotherapy approach.
Eliza operated by utilizing an extensive set of pre-programmed responses. Upon receiving input, it would sift through these responses to find the most relevant and appropriate one, often modifying it slightly, according to the input, before outputting it.
This early program can be viewed as a rudimentary form of natural language processing, akin to contemporary systems like ChatGPT, but with a much much more simpler data set, since ChatGPT have scrapped the whole today’s internet. A more fitting analogy might be Microsoft Word’s Clippy, which similarly provided predetermined responses based on user input.
You can chat with Eliza here : https://web.njit.edu/~ronkowit/eliza.html
And you can see a part of it’s original source code, paper-printed (those where the days) below :
Comments