The first chatbot mimicked psychiatry as early as the 1960s and succeeded

World The first chatbot is brought to life. Eliza returned to life just before Christmas 2024, after about 60 years of break.

The researchers found and returned Eliza’s code, which was thought to have disappeared. Chatbot works just fine now.

The two “software archeologists” of MIT university praise Eliza. Some thought Eliza was a very good talker.

Now Eliza, who has taken a break, can again ask a person, “Could you tell me why you are not happy?”

Discussion robot Elizan was developed in the 1960s Joseph Weizenbaum.

It got its name from the play Pygmalion. There’s the main character Eliza Doolittle They taught how to speak like a upper -class British woman. The same person is better known for the musical My Fair Lady.

Researchers in artificial intelligence say that Eliza had a great impact on today’s artificial intelligence. It was once a kind of language model with which the user interacted with.

The original Eliza reported Eliza Archeology Project. Of course, a lot of versions developed from Eliza have been seen in the world, says New Scientist.

“Software archeologists” write about their discoveries In the Arxiv web service. The writing is dated January 2025.

In the church A manuscript called “Doctor” was prepared for the original program. It refers to a psychotherapist.

Doctor was programmed to discuss in the same way as a psychotherapist. Weizenbaum wanted to explore how machines and people could discuss.

For example, Eliza could start, “Tell me about your problems.”

If the user fed the phrase, “The men are all the same,” may have been asked, “How?”

Weizenbaum wrote Eliza in a programming language he developed, which is no longer used. Its name was Mad-Slip.

The program was soon copied to the LISP language. It became more common in the early years of the Internet. At the same time, Eliza’s LISP version became more common, and the original Eliza began to forget.

For decades As they progressed, many thought Eliza’s original 420-line code had disappeared.

In 2021 Jeff ShragerCognitive scientist at Stanford University, and Myles CrowleyHowever, MIT’s archivist, found Eliza’s program. It was in the retarded papers of Weizenbaum.

“I am interested in how early artificial intelligence pioneers thought,” Shrager said about his discovery online service Live Science for an email. He is the second author of the study published in Arxive.

“Eliza was and is good or bad for early artificial intelligence,” says Shrager.

Finders At first, they did not know if the code was found.

Eliza’s revival was not easy. The revitalists had to develop its style computer that would have used Eliza in the 1960s.

Eliza’s returners got the program in order in December 2024, after about 60 years of break.

Although Eliza worked very well, its rescuers also found an error in the code. The finders decided not to fix it.

“It would ruin authenticity,” Shrager explains.

Eliza There was something completely new in the 1960s. Its abilities exceeded all expectations. During the trial, some users thought they were talking to the right therapist on the line, that is, with a person.

Otherwise, human emotions were reflected in Eliza. For example, Weizenbaum’s own secretary did so, even though he knew Eliza was a machine.

Eliaaa Of course, you can’t compare to current LLM language models like chatgpt.

But Eliza gets a lot of forgiveness because its program started to be done as early as 1964, says Wikipedia. Eliza also became some kind of pop phenomenon.

“Eliza stays in the conversation for a while,” praises its style David Berry. He is a professor of digital humanities at Sussex University in England.

One thing Eliza controlled better than modern chatbots. It could listen, and can still.

Discussing artificial intelligence LLM models often try to supplement and guess your phrases you enter on the machine.

Eliza was programmed to urge the user to continue the discussion.

“It’s more” chat “than any chatbot after that,” Shrager praises.