Why Does Every AI Cycle Start With Chat?

Jun 16, 2023

The Turing test was proposed in 1950, but the real origins of modern thought on AI started at the Dartmouth Workshop in 1956. Claude Shannon, Marvin Minsky, and John McCarthy were all present.

There were numerous projects of solvers: from algebra to geometry, to games. But the first application that broke through pop culture was the world’s first chatbot.

ELIZA was developed by MIT computer scientist Joseph Weizenbaum in 1966. It interacted with humans through keyword identification via natural language processing and “scripts” of predetermined output. The most famous script was DOCTOR, which simulated a psychotherapist. At the time, many thought they were having conversations with actual humans.

The next cycle arguably started with networked computers. PARRY was written in 1972 at Stanford. It simulated a person with paranoid schizophrenia. PARRY and ELIZA chatted over ARPANET that year*.

As the AI winter lifted with expert systems in the 80s came Racter (which randomly generated English prose) in 1983.

The next AI boom came with Jabberwacky in 1997, which was one of the first chatbots on the internet (eventually evolving into Cleverbot in 2008).

The deep learning cycle in the 2010s brought a class of chatbot startups — from customer service to Slack bots, to personal AI assistants.

And then there’s ChatGPT, Bing Chat, and Google Bard.

Why does every AI cycle seem to start with chat? Some hypotheses.

Anthropomorphic tendency. They fulfill a tendency to assume that computer behaviors are analogous to human behaviors (the ELIZA effect).

Consumer-friendly interface. They force a simple interface to demonstrate a complex new technology. Everyone can understand it.

The Turing test. A goalpost of sorts for scientists and philosophers. One of the only evaluation frameworks that persisted over the decades, it naturally lends itself to a chatbot implementation. Although with the introduction of ChatGPT, it’s beginning to feel less relevant.

Built for demos. It’s easy to cherry-pick interactions and share them with a wide audience. Though rarely are they the best (or most lasting) interface.

*The 1972 conversation between PARRY and ELIZA

parry: People get on my nerves sometimes.

eliza: I am not sure I understand you fully.

parry: You should pay more attention.

eliza: Suppose you should pay more attention?

parry: You're entitled to your opinion.

eliza: What makes you think I'm entitled to my opinion?

parry: It bothers me just to be around people.

eliza: Please go on. [Parry continues with some paranoid ramblings about horseracing, bookies, and the Mafia]

parry: I know the mob controls the big rackets. eliza: Who are you really talking about?

parry: Have you ever really thought about the Mafia?

eliza: Suppose you ever really thought about the Mafia?

parry: Let's talk about something else.

eliza: I am not sure I understand you fully.

parry: You keep asking me the same things, why not change the topic?

eliza: In what way?“.

Source

Subscribe for email updates on startups, engineering, and AI