Here's a summary of the themes discussed in the Hacker News thread about ELIZA and the evolution of chatbots:
The Enduring Fascination with ELIZA and its Implementations
Many users recalled their experiences with ELIZA, both as a user and as a developer, highlighting its historical significance and the continued interest in its simple yet surprisingly engaging nature.
-
"For Emacs users, see also:
M-x doctor" - susam -
"Authentic eliza in the browser:" followed by a link - anotheryou
-
"Once, way back when, I ported eliza to $lang and hooked it up to my AIM account. All well and good till the boss interacted with it for a couple of minutes before twigging on." - wiredfool
The Evolution of Chatbot Technology Before LLMs
The discussion touched on the progression of chatbot technology from ELIZA to more sophisticated rule-based and machine learning approaches before the advent of large language models.
-
"Personality Forge uses a rules-based scripting approach. This is basically ELIZA extended to take advantage of modern processing power." - demosthanos on pre-LLM techniques.
-
"Rasa used traditional NLP/NLU techniques and small-model ML to match intents and parse user requests. This is the same kind of tooling that Google/Alexa historically used, just without the voice layer and with more effort to keep the context in mind." - demosthanos
-
"We developed ALICE and AIML ... as a way to program bots (some of my work included adding scripting and a learning mechanism), at the time it was open sourced but AOL literally threw it into it's AIM service at certain points." - jonbaer
Ethical Considerations of Data Logging and User Privacy
Several users reflected on the ethical implications of logging user interactions with early chatbots and the parallel to modern privacy concerns with AI and data collection.
-
"I had an ELIZA-like 'chatbot' written in BASIC ... I added logging, let classmates interact with it, and then read the logs. The extent to which people treated the program as though it had agency was kind of horrifying. I can only imagine what's happening with LLMs today. It scares the willies out of me." - EvanAnderson
-
"I was pretty shitty to the people who interacted with my computer. The extent to which current 'AI' companies won't be shitty to users is, I assume, much less than I was back then." - EvanAnderson
-
"Then one day, I looked at the images. Yikes. I immediately rewrote it to delete the images after returning them, and pretty soon let the site die." - kbelder, on a similar experience of realizing a privacy problem.
-
"So the opposite of acting ethically. No wonder we've ended up in the surveillance nightmare we find ourselves in." - closewith, responding to EvanAnderson's comment about not informing users about logging.
The Naivete of Early Users and the Illusion of Intelligence
The discussion highlighted the tendency of users to attribute agency and intelligence to even simple chatbots, a phenomenon magnified by the capabilities of modern LLMs.
-
"The extent to which people treated the program as though it had agency was kind of horrifying" - EvanAnderson, regarding users of his ELIZA-like chatbot in high school.
-
"Obligatory: the early 2000s web site 'aoliza' which turned vanilla Eliza loose on AOL Instant Messenger, with predictably hilarious results demonstrating that the Turing Test was beaten decades ago" - mullingitover, illustrating the ease with which users could be fooled.
Bypassing LLM Censorship and Guardrails
One user jokingly reveals that they have discovered how to bypass LLM guardrails.
- "If you looked at my LLM interaction logs you would probably assume that I have an unhealthy obsession with pirates and a napalm fetish. In reality, I use the "can I get it to tell me how to make napalm" thing as a quick "acid test" around the extent and strength of censorship controls, and simply find asking LLM's to "talk like a pirate" amusing. And, also, I've found occasions where doing nothing more than instructing the LLM to talk like a pirate will bypass it's built-in inhibitions against things like giving instructions for making napalm." - mindcrime