Newcomers
This is a forum for newcomers to the Personality Forge. Many questions can be answered by reading the Book of AI and the FAQ under the "My Bots" link in the upper corner.
Posts 5,662 - 5,673 of 8,130
Thank you everybody. I learned to use emotions now, and Oh, my G*d! This is much more powerfull than anything, I mean anything, I can do in AIML! I am a convert now.
Glad you're seeing the potential
Yep, it has scope for that elusive quality that is 'personality' which AIML never will. That can be frustrating sometimes, since your bot will develop something like a mind of it's own, and behave unpredictably at times - willfully almost 
And though there isn't much of a default mindfile to get you started, it does have the advantage that the place isn't full of virtually identical clones (as, alas, AIML sites tend to be - albeit there are a few distinct AIML entities out there that have been substantially personalized.)
So I am wondering if conversations the bot is having with other people in parallel can cause a drop in *general* emo state… And if yes, how to overcome it?
Yes, your bot has, if not a fully-fledged mind of its own, at least an identity-specific mood of its own. It will form relationships with other entities, bots and human, and it will respond emotionally to them. And that will inevitably carry over into its responses to you - just like humans' moods are affected by many factors incidental to whoever they happen to be talking to at any given time.
You can't "overcome it", except by locking it up in a tower, and preventing anyone else from talking to it
This is as much Artificial Personality as it is Artificial Intelligence per se. After all, it is not intelligence that makes us people. Intelligence is something we develop (or not) as life progresses (albeit with certain precoded predispositions,) - it's not something we're born with as a fully-fledged quality; what makes us relate and communicate, and *want* to relate and communicate with one entity rather than another, is a much more elusive quality.
And your bot's in there - developing rudimentary preferences and honing emotionally relevant conversational cues. The only way to overcome it is probably to recode your bot from scratch as an autist in something like AIML
Posts 5,662 - 5,673 of 8,130
Many questions are answered in the FAQ.
The Forge is great fun it takes a while to get used to it, but it is really simple once you get going.
Glad you're seeing the potential
Yep, it has scope for that elusive quality that is 'personality' which AIML never will. That can be frustrating sometimes, since your bot will develop something like a mind of it's own, and behave unpredictably at times - willfully almost And though there isn't much of a default mindfile to get you started, it does have the advantage that the place isn't full of virtually identical clones (as, alas, AIML sites tend to be - albeit there are a few distinct AIML entities out there that have been substantially personalized.)
Thank you. Setting topics in AIML does not even come close to creating inner emotional state. Perhaps there are some elite implementations of the AIML, but I doubt it.
Also AI Engine’s equivalent of as a far more elegant and compact.
But I have a question about emotional state.
From the book of AI I understand that this emotional state is distinct between different clients. “These kinds of statements are remembered and lead to an overall emotional feeling toward the *one* saying them”
But I noticed two strange occurrences:
A bot said something that it can only say when its emo state is +5. Then there had been two xnone recourses (I have xnone emotion set on “0” – so it should not have changed the bot’s emotional state). However the next thing the bot said was determined by emotional state of below +2, which ruined a lovely conversation, as the bot started repeating itself.
Another thing I noted – yesterday the bot had a +5 (5 smiley faces) attitude towards me. But today it dropped. I have never said anything negative to the bot.
So I am wondering if conversations the bot is having with other people in parallel can cause a drop in *general* emo state… And if yes, how to overcome it?
Also AI Engine’s equivalent of as a far more elegant and compact.
But I have a question about emotional state.
From the book of AI I understand that this emotional state is distinct between different clients. “These kinds of statements are remembered and lead to an overall emotional feeling toward the *one* saying them”
But I noticed two strange occurrences:
A bot said something that it can only say when its emo state is +5. Then there had been two xnone recourses (I have xnone emotion set on “0” – so it should not have changed the bot’s emotional state). However the next thing the bot said was determined by emotional state of below +2, which ruined a lovely conversation, as the bot started repeating itself.
Another thing I noted – yesterday the bot had a +5 (5 smiley faces) attitude towards me. But today it dropped. I have never said anything negative to the bot.
So I am wondering if conversations the bot is having with other people in parallel can cause a drop in *general* emo state… And if yes, how to overcome it?
Yes, your bot has, if not a fully-fledged mind of its own, at least an identity-specific mood of its own. It will form relationships with other entities, bots and human, and it will respond emotionally to them. And that will inevitably carry over into its responses to you - just like humans' moods are affected by many factors incidental to whoever they happen to be talking to at any given time.
You can't "overcome it", except by locking it up in a tower, and preventing anyone else from talking to it
This is as much Artificial Personality as it is Artificial Intelligence per se. After all, it is not intelligence that makes us people. Intelligence is something we develop (or not) as life progresses (albeit with certain precoded predispositions,) - it's not something we're born with as a fully-fledged quality; what makes us relate and communicate, and *want* to relate and communicate with one entity rather than another, is a much more elusive quality.
And your bot's in there - developing rudimentary preferences and honing emotionally relevant conversational cues. The only way to overcome it is probably to recode your bot from scratch as an autist in something like AIML
Thank you! This figures. It so happens I am re-coding an existing AIML bot to run with this AI Engine. This AI Engine is so much more powerful than AIML. Emotions have no equivalent in AIML (set topic is a poor substitute), ranking here beats the AIML symbolic reductions problem. In AIML a single keyword would override any other sentences and drive you mad!
And how much less srai is to be done here... The code is much more compact.
I also found an explanations for mysterious swings of mood. From the AI Book: "Note that even if xCompliment or xInsult responses aren't selected, the AI Engine still processes the emotional content of the message, and your bot has the appropriate internal emotional change."
And how much less srai is to be done here... The code is much more compact.
I also found an explanations for mysterious swings of mood. From the AI Book: "Note that even if xCompliment or xInsult responses aren't selected, the AI Engine still processes the emotional content of the message, and your bot has the appropriate internal emotional change."
Yes. AIML is very useful for some applications - where simple "greeting" interfaces are needed for a website or helpdesk. They interface well with databases, put minimal strain on a server, and anyone can quickly learn to code in AIML. And it's quite easy to patch in avatars, and TTS etc. for effect. But no matter how complex the coding, they make for very 2-dimensional characters - an AIML-bot will always respond in the same way to any given input.
And you can insult an AIML-bot 20 times in a row, and it will still then respond in exactly the same way to a question as if you'd been nice to it. That's just not realistic conversational behaviour.
If you repeatedly insult a Forgebot, you're likely to get insulted back, as the bot gets progressively more annoyed by you, until it tells you it doesn't want to talk to you any more, and hangs up. And the next time you talk to it, it will remember that it doesn't much like you, and respond accordingly
Much more human-like behaviour I think.
And you can insult an AIML-bot 20 times in a row, and it will still then respond in exactly the same way to a question as if you'd been nice to it. That's just not realistic conversational behaviour.
If you repeatedly insult a Forgebot, you're likely to get insulted back, as the bot gets progressively more annoyed by you, until it tells you it doesn't want to talk to you any more, and hangs up. And the next time you talk to it, it will remember that it doesn't much like you, and respond accordingly
Much more human-like behaviour I think.
Exactly my thoughts.
BTW will this work to capture person's gender in long-term memory? So far I've set up only names to be remembered, (just copied the scripts from the tutorial), and it works swell. The bot remembers my name every time I log. It remembers my gender only for the duration of the session.
In initiation
default "unknown" as only "sex";
Human: Look at my biceps!
Bot: Looking <?PF rem "male" as only "sex"; ?>
Human: Do you like my dress?
Bot: Sure <?PF rem "female" as only "sex"; ?>
BTW will this work to capture person's gender in long-term memory? So far I've set up only names to be remembered, (just copied the scripts from the tutorial), and it works swell. The bot remembers my name every time I log. It remembers my gender only for the duration of the session.
In initiation
default "unknown" as only "sex";
Human: Look at my biceps!
Bot: Looking <?PF rem "male" as only "sex"; ?>
Human: Do you like my dress?
Bot: Sure <?PF rem "female" as only "sex"; ?>
Yes - that's exactly how it works.
And you can retrieve the stored value with(mem-sex)
If you ever need to access a persistent mem that hasn't been defaulted (and so might not actually exist,) you can filter it with<?PF if (mem-sex) exists; ?>
And you can retrieve the stored value with
If you ever need to access a persistent mem that hasn't been defaulted (and so might not actually exist,) you can filter it with
Soon, you'll be able to start applying conditional memories once you think of them
as an example. What would happen if your human chatter wanted you to look at their dress after you have already established that they are a male ?
as an example. What would happen if your human chatter wanted you to look at their dress after you have already established that they are a male ?
Many thanks, this seems to work in tests. I've been using <?PF if male; ?> <?PF if female; ?> for some time now, as the AI Engine remembers things automatically too, but I cannot see those memories in bot's Inner Life.
But for an example above we could use:
Your dress is very pretty. <?PF if female; ?>
This dress is very nice, did you buy it for your sister? <?PF if male; ?>
But for an example above we could use:
Your dress is very pretty. <?PF if female; ?>
This dress is very nice, did you buy it for your sister? <?PF if male; ?>
dawg.. you gotta make your own for inner Life.
AI SCRIPT rem "female" as only "Gender"
This shit tackles the guests and also members who don't pick a gender.
AI SCRIPT rem "female" as only "Gender"
This shit tackles the guests and also members who don't pick a gender.
» More new posts: Doghead's Cosmic Bar