Google holds 14% of Anthropic equity, with a total investment of over US$3 billion
641
English
English
After more than 60 years in the dust, the world's first chat robot ELIZA has finally seen the light of day. A team of AI historians and computer scientists have managed to get the AI pioneer running again on its original operating system, opening a window into computing history.
The breakthrough began in 2021, when researchers discovered Joseph Weissenbaum's original ELIZA code in MIT archives. This rare find contains the complete DOCTOR script - the core program that enables ELIZA to imitate Rogers' psychotherapy techniques, which has previously only existed in modern programming language recreations.
The process of resurrecting ELIZA was full of challenges. The team was required to work on 2,600 lines of barely documented code and run the program on a simulated IBM 7094 computer. During the repair process, researchers accidentally discovered a previously unknown "teaching mode" that allows users to customize the behavior of ELIZA by entering "+". To maintain historical authenticity, the team even retained a special flaw in the original program that caused it to crash when encountering numbers.
The long-established ELIZA chatbot presents her unique dialogue on the CTSS system. The interaction demonstrates typical therapeutic conversation techniques, in which ELIZA receives the user's statements and feeds them back as questions.
The rebirth of ELIZA is not only an important moment in computer history, but also triggers in-depth thinking on the nature of modern AI. It raises a fundamental question: Does AI need to completely replicate the human thought process to achieve intelligence? This question is still of great significance in today's era of large language models such as ChatGPT.
Researchers pointed out that ELIZA is the first robot to embody the Turing test, pioneering human-machine dialogue. Today, in the context of the rapid development of AI, this concept is more inspiring than ever. The restored ELIZA code has been released on Github for anyone interested in computer history to explore this unique technological legacy.
