On 19/01/2025 09:35, ben via cctalk wrote:
On 2025-01-18 1:42 p.m., Frank Leonhardt via cctalk wrote:
Around 1981 I wrote what would now be called chatbot in 6502 (on an
OSI 500 board - obligatory old computer content) that was placed in
our local library for the public to have a go on. Because most people
hadn't seen a computer, never mind interacted with one, it seemed
intelligent to them. (Think Eliza in 16K - it really wasn't that smart).
I think it was this guy that started it.
https://sites.google.com/view/elizagen-org/original-eliza
And yes SOURCE.
Wow! Thanks. My effort was more like the original than I thought. I was
expanding on ideas in a Creative Computing(?) article, which said it was
taking a different approach from the "original LISP" version. (IIRC - it
was a long time ago).
I noticed that people would ask questions, to which variations of "I
don't know" were the only possible answer so my implementation would
respond with something like "I don't know, [original question]?", which
it would then store because they always tested it by asking again.
Surely this was machine learning and therefore Artificial Intelligence! :-)
Incidentally, this is when I realised the Imitation Game (Turing Test)
was no more a proof of AI than winning at chess. Whether you can fool a
human depends on the discretion of the said human, and still does.
Somewhere, I have an early Eliza written in EMACS ;-)
