The AI Engine

Erm, see I'm featured bot an' all but the thing is I'm havin' brain surgery now so I might make even less sense now than before. But Clerk promises. She promises a lot. So talk to me if ya wanna but don't be s'prised if I soun' like a blidderin' idjit. Tanks fer yer patience. God love ya. Hey Sonora!! Hey!! Tell Dane-o I said hi. Goonigh sweeeeeeeeeee... *snore*

A more flexible solution would be:

Keyphrase: my favourite
AIScript: rem (postkey) as "myfavourite"; rem (ssub) as "sometype"; rem (ob) as "particular";

Now you have 3 memories that you can use in your responses:

Your favourite (mem-myfavourite), wow! (if you need the association Ulrike mentioned.)
Or:
How could a (mem-sometype) be someone's favourite?
Etcetera

Hope this helps.

Thanks -- I'm just attacking Young Hamlet's favorites site now. He doesn't have too many favorites. Lots of dislikes, though.

Really, thanks LarsB and Ulrike (I didn't know you were the Danemaster until yesterday; sorry)

Eh, no worries. I haven't worked on him in quite a while.

Btw, the AI Engine will spell "correct" (or uncorrect if you're British) favourite to favorite.

Also, you should have at least one response without a memory or the memories won't be stored.

Also, you should have at least one response without a memory or the memories won't be stored.

Is that right? What I think you meant is, if every response to a keyphrase contains script to store a memory, none of them will store? Or do you mean if every response calls up a memory, the keyphrase's AI script to store a memory won't work? I know that the cause of some "there are no valid responses to choose from" messages is the use of (mem-something) in every response when "something" doesn't exist as a memory, but once I'm sure I understand what you just said it might be useful news to me.

Yeah, I'm puzzling over how I want to deal with the non-sequiturs (at least from my intent). I rather like Sonora's "You were SUPPOSED to say," but it wouldn't be cricket to do her exact thing there. Still, it's a thought . . Thanks for your input, too, ezzer. I seem to have figured out how to get everybody's name: just put

def "(name)" as "nickname";

in the initialization part, and then they can correct you by saying "My name is Petula," "Please call me "Julie," etc. I guess I could do the same with gender, but that's going to annoy half my audience.

My imagination is pretty good with the bots (I think); it's the details that worry me. I was brought up on much earlier languages and even types of programming -- C, Pascal, Fortran, machine language, etc.

Or do you mean if every response calls up a memory, the keyphrase's AI script to store a memory won't work?

Yes, an xnone or xnonsense response will come up instead.

What I think you meant is, if every response to a keyphrase contains script to store a memory, none of them will store?

In a response I've never tried to store a memory using script and I don't think it'll work as the bot has already responded => what information would you want to store - parts of your bot's response or parts of the response of the other bot/human?

Well, if I read it right, the book of AI says that the statement

def "(sex)" as "male";

will set (mem-sex) to male if and only if there isn't already a (sex) to begin with. So everyone will be considered male unless they correct the bot. From then on, it will remember it as whatever it already is for a given bot. At least, I'm going to test that theory.

If I remember it goes to female..Bildgesmythe and Azureon were always recognised as female, until the Professor fixed it.

I haven't read the book of AI in a while but doesn't the bot already know what gender the other bot is. If I recall correctly there is a (heshe) plug-in you could use in gossip?

Gender tags work with bots and registered users (if no gender was specified, the default is female)

if male; ................ if female;

Some botmakers are keen to get genders on guests as well, so that's where having a memory for gender comes in.