This page was generated programmatically. To view the article in its initial location, you can follow the link below:
https://www.livescience.com/technology/eliza-the-worlds-1st-chatbot-was-just-resurrected-from-60-year-old-computer-code
and if you wish to have this article removed from our website, please reach out to us
Researchers have recently revived “ELIZA,” the first-ever chatbot, from forgotten computer code — and it still functions remarkably well.
Using dusty printouts from MIT archives, these “software archaeologists” unearthed obsolete code that had been missing for six decades and revived it.
ELIZA was created in the 1960s by MIT instructor Joseph Weizenbaum, named after Eliza Doolittle, the central character of the play “Pygmalion,” who was instructed how to speak like an elite British lady.
As a linguistic model that users could engage with, ELIZA significantly influenced modern artificial intelligence (AI), the researchers noted in a document shared on the preprint repository arXiv on Sunday (Jan. 12). The “DOCTOR” script crafted for ELIZA was designed to reply to queries as a therapist would. For instance, ELIZA might say, “Please share your problem.” If the user input was “Men are all alike,” the response from the program would be, “In what way.”
Weizenbaum developed ELIZA in a now-obsolete programming language he created, known as Michigan Algorithm Decoder Symmetric List Processor (MAD-SLIP), but it was swiftly adapted into the Lisp language. With the rise of the early internet, the Lisp variation of ELIZA became extremely popular, rendering the original version outdated.
Related: Google’s AI suggests users add glue to their pizza, eat rocks, and produce chlorine gas
Experts believed the original 420-line ELIZA code was lost until 2021, when co-author of the study Jeff Shrager, a cognitive scientist from Stanford University, along with Myles Crowley, an archivist at MIT, found it within Weizenbaum’s documents.
“I have a specific interest in how early AI innovators perceived their field,” Shrager shared with Live Science via email. “Having computer scientists’ code is as close to accessing a record of their thoughts; as ELIZA was — and continues to be, for better or worse — a landmark of early AI, I want to understand what he was pondering.” However, the reasons behind the team wanting to activate ELIZA are more intricate, he clarified.
“From a technical standpoint, we were uncertain if the code we unearthed — the sole version ever located — would indeed function,” Shrager mentioned. Thus, they acknowledged the need to test it.
Reviving ELIZA
Reintroducing ELIZA was far from simple. It necessitated the team to clean and troubleshoot the code and to construct an emulator that would simulate the type of computer that would have operated ELIZA in the 1960s. After restoring the code, the team successfully got ELIZA running — for the first time in six decades — on Dec. 21.
“By executing it, we demonstrated that this was, undeniably, a segment of the actual ELIZA lineage and that it not only functioned but functioned remarkably well,” Shrager stated.
However, the team also identified a flaw in the code, which they decided against fixing. “It would compromise the authenticity of the artifact,” Shrager clarified, “similar to amending a mis-stroke in the original Mona Lisa.” The program malfunctions if a user inputs a numeral, such as “You are 999 today,” as noted in their study.
Although it was intended to serve as a research platform for human-computer interaction, “ELIZA was such a novelty at the time that its ‘chatbotness’ overshadowed its research intentions,” Shrager elaborated.
Related: 32 occasions when artificial intelligence made catastrophic errors
This legacy persists today, with ELIZA often juxtaposed with contemporary large-language models (LLMs) and other forms of artificial intelligence.
Even though it does not measure up to the capabilities of modern LLMs like ChatGPT, “ELIZA is genuinely remarkable considering it was developed in 1965,” David Berry, a professor of digital humanities at the University of Sussex in the U.K. and co-author of the paper, remarked to Live Science in an email. “It can sustain a conversation for a period.”
One aspect where ELIZA outperformed modern chatbots, according to Shrager, is in listening. Current LLMs primarily attempt to finish your sentences, while ELIZA was programmed to encourage users to continue the dialogue. “That aligns more with what ‘chatting’ entails than any purposeful chatbot since,” Shrager commented.
“Reviving ELIZA, one of the most — if not the most — iconic chatbots in history, broadens people’s awareness of the history that is at risk of being lost,” Berry expressed. Due to the forward-looking nature of computer science, many professionals tend to deem its history as irrelevant and neglect to preserve it.
Berry, however, maintains that the history of computing is also a part of cultural heritage.
“We must strive more diligently as a society to preserve these remnants of the early days of computation,” Berry urged, “because if we fail, we will have lost the digital equivalents of the Mona Lisa, Michelangelo’s David, or the Acropolis.”
This page was generated programmatically. To view the article in its initial location, you can follow the link below:
https://www.livescience.com/technology/eliza-the-worlds-1st-chatbot-was-just-resurrected-from-60-year-old-computer-code
and if you wish to have this article removed from our website, please reach out to us
