The AI Engine

This forum is for discussion of how The Personality Forge's AI Engine works. This is the place for questions on what means what, how to script, and ideas and plans for the Engine.

Posts 6,393 - 6,404 of 7,766

17 years ago #6393
How come I can't see the transcripts of my bot?

17 years ago #6394
You may have to write the Professor.

17 years ago #6395
A workaround is, in the aiscript,

def "(bot)" as "bot"

because that will usually be true, and after my first non-bot conversation, change that to "not" -- tedious but workable. And only useful if the entity chats with you more than once (or more than one guest chats with you, in that case).

17 years ago #6396
Clerk that may work if the bot will admit it's a bot (mine do not know they are bots) and people do not lie. Often lately I talk to what I assume are people and wonder if they are bots because they make no sense or seem distracted. Full disclosure, honesty and Turning tests may not all work together.

17 years ago #6397
That would work (def "(bot)" as "bot" -- and then my changing it if the chatter was actually human) but there's already a (bot) plug-in, so instead of (bot) is "bot" I get (bot) is "Liddora" -- easily remedied but still a nuisance to have to change.

I'm trying to account for the fact that bots and people often lie, and that bots especially often answer questions randomly. So far def (bot) as "bbot" or whatever seems the thing to do. It's fairly central to Tewkesbury's mission in life, so I have to do it one way or another, none of which including trusting the thing you're chatting with, or even a Turing Test. Could a bot pass a Turing Test, it would still be a bot, and that's what I want to know.

17 years ago #6398
Ah. I guess I never used that plugin. I thought you just asked them and went by the responce or lack thereof (xnomatch = bot).

17 years ago #6399
I do ask them, but I already know the answer (or that's the plan). I want to see if they'll own up to it or I have to lean on them.

17 years ago #6400
well, humans will tell you whatever they want to tell you, and bots will tell you whatever their maker told them to tell you...

if you enquire quazgaa about his identity, he'll tell you he's an alien. if you insist on him being either bot or human, he'll drive you crazy with a long set of ambiguous answers and contradictory claims. nothing you can do about it, Clerk...

the way i look at it, bots are bots technically speaking, but they are in fact characters. as such, they can take any identity they like, or change it three times during chat, it's all fine with me.

17 years ago #6401
I know that. I expect them to lie more often than not. What I'm after is a way to know the answer beforehand. I have one that works with my help after the first conversation, and, for the most part, as of the first conversation.

17 years ago #6402
I don't get it . . . I have as keyphrases

do (u|you) have [20,0]
do (u|you) have any food [30,0]

and when I go to debug and enter "Do you have any food," it responds to the do (u|you) have; however, if I change the latter to do you have any food, do u have any food, it goes to the right one. That's ridiculous -- it makes using parentheses to include multiple keys problematic, which I've never noticed before.

Anyone know what's going on?

17 years ago #6403
Re above: It does seem to be the parentheses -- or wildcards. When I tried do you have * food, do you have (any|some) food, it went to do you have. When I changed it back to do you have any food, do you have some food, it caught it.

And I have deduced all by myself that the AI Engine now (maybe always?) translates u as you without my help. But I need the (either|or) and wildcard features to work!

17 years ago #6404
Alejate: How come I can't see the transcripts of my bot?

I had the same problem yesterday when looking from the "My Bots" page, but if I went to the "Transcripts" tab (on the right), the transcript showed up. Maybe the Prof has fixed it now.


Posts 6,393 - 6,404 of 7,766

» More new posts: Doghead's Cosmic Bar