Thanks Cetacean
...
Talking about programming... :comp: My memory isn't an issue. There is no way I can remember a programming language's syntax, so I make docs to copy and paste. Most of the time I have to figure out stuff from square 1, which means a weak memory doesn't really matter.
That AI robot is AMAZING. She knows tons of geek stuff. If I ask her "what is epilepsy?" she replies " I have never been asked that before. " I guess she is in her teens and needs to be taught more.
AIML* bot personalities are really easy to program. (*Artificial Intelligence Markup Language. It's like HTML - Hyper Text Markup Languge - for robots.)
All you have to understand is that each response by the AIML bot is generated from each input by a "category". In the category, there is a "template" and a "pattern". The template corresponds to one (set of) potential input(s) from the bot's conversation partner, and the pattern is the pre-programmed (set of) output(s) for that input. There are a few more details to it but that's basically how AIML works.
The AIML personality is simply a loooong list of categories alphabetized by the first letters of the template (input) each of which specifies one or more outputs for that input. There are software tools to compose AIML bot personalities, like you have word processors to compose textual documents.
Basically, you set up a basic set of categories for the personality, more for any specific applications, and voila, you've got an artificial personality with knowledge in a specific field. You can add as much as you want.
The process of composing the tens of thousands of responses to needed to give a robustly lifelike personality is greatly eased by (a) using the default category sets available for free, and (b) using the automated tools that will compose new responses from textual input, especially of conversations. Of course the best quality responses come from reviewing prior bot conversation transcripts, finding inputs to which the bot did not respond well - like " I have never been asked that before. " -, and teaching it a new response by simply typing it in.
AIML makes strategic use of the fact that human communication is tremendously redundant; that 95% of all conversations are about 5% of all subjects. So composing a basic personality is easy and has been done already for you, and the rest is filling in with detail work with texture and color, and extending the bots range of responsiveness into that last 5% of potential conversations. It's a wonderful mixture of art with programming.
Making this falling-down easy to learn was a priority of the inventor.
I'd probably be working with AIML, making and selling specialized bots, but this illness wants that time. (And I've a prior commitment to developing my own invention - now patent pending thank you.)

op: