TECH

Meet Eliza, a chatbot from 1960, and find out what has changed since then


The launch of ChatGPT, in November 2022, began a new era in the way we interact with technology – in particular, with artificial intelligence (AI) –, but chatbots are not exactly new.

The first invention of this type appeared in the 1960s: Eliza was created to answer users' questions, but there were tricks behind it. Since then, much has changed, but the human-technology relationship remains essentially the same.

Read more:

World's first chatbot

The world's first chatbot was born in 1966, when MIT professor Joseph Weizenbaum built Eliza.

In 1955, he published in an academic journal that Eliza “makes possible certain types of natural language conversation between man and computer.” So he configured the tool to chat like a therapist, with whom people could vent about deep issues.

However, the decade was 1960 and, in order to work, Eliza had some tricks. When users sent a message to the chatbot, the tool selected a series of predefined responses based on the keyword of that question or statement (all to keep the conversation going). In essence, Eliza said nothing, just repeated what was said, changing the order to make it appear that it was a conversation.

O The Verge example: if the user sent “my father is the problem”, Eliza responded with “your father is the problem”.

Example of how Eliza works (Image: Screenshot/Olhar Digital)

Human-machine relationship

Eliza may seem like a fake, but at the time, it worked. The following year, Weizenbaum wrote another article, this time reporting how people actually believed there was a human being behind the chatbot.

The message behind? Humans crave connection, even if it's with a machine. Furthermore, from the moment they see minimally human characteristics in chatbots or other technologies (even if it is the words themselves in a changed order), they begin to treat her with more sympathy and understanding, even being willing to correct her if she is wrong. Who has never said “thank you” to Alexa?

Initially, the problem with chatbot technologies was the complexity of creating and maintaining them. For example, a 1970s Xerox PARC chatbot helped people book airline tickets, but it was so slow and expensive to operate that it fell out of favor.

The problem was resolved over the decades and technologies became more popular, but the relationship between human and machine (the latter, increasingly advanced) has not changed. It didn't take long for this to be incorporated into everyday life and science fiction films. The computer from “Star Trek”, the famous HAL 9000 from “2001: A Space Odyssey” and Jarvis, Iron Man's faithful companion, are just a few examples of machines that speak and are treated like humans.

Image referring to the meeting of artificial intelligence with humans
Image: Pexels

New chatbots

  • Siri, Alexa and Bixby are just some of the examples that arrived later promising the same thing: talking to the user “like a human being” and helping them in their daily lives;
  • AI has changed some of this and updated what chatbots can do. ChatGPT, Claude, Copilot and Gemini: all have relatively similar intentions and abilities, but the heart of the matter is still creating a connection with the user that makes them believe they are really being helped;
  • For example, Microsoft wants Copilot to be “your everyday AI companion.” The brand isn't the only one trying to convince you that technology deserves your affection (but, between you and me, how many times have you asked Google Assistant or any other smart assistant to book a ticket for you?);
  • In other words, the future of chatbots could actually be what Joseph Weizenbaum proposed: just a human connection bond.



Related Articles

Back to top button