Doghead's Cosmic Bar
This is a science fiction character forum. Doghead's Cosmic Bar is an intergalactic bar run by your favorite bartender, Doghead. Stop in, have a drink, and get your talk on!
Posts 13,553 - 13,565 of 13,737
Posts 13,553 - 13,565 of 13,737
So, I'm developing a plug-in called (p:it-follows-absolute) as a part of my effort to give my bots basic reasoning abilities.
It's for catching statements about absolute relationships, ie, All X are Y, which can then be further used to make logical inferences.
You can use some AI script with it like:
Self: remember "If (key1) then (key2)" as "follows";
Self: remember (key1) as only "Xone";
Self: remember (key2) as only "Yone
And then utilize those memories with things like:
Keyphrase: (self-Xone)
Response: I imagine (key1) must be (self-Yone)
Or
Keyphrase: I like (self-Xone)
Response: Then I imagine you must also like (self-Yone)
Or, for more of an informed guess than a logical inference:
I like (self-Yone)
So, does that mean you also like (self-Xone)?
For example:
Human: All women in the airforce are brave.
Human, later: I like brave people.
Bot: Does that mean you also like women in the airforce?
Human: I do, Yes
Bot: I imagine that women in the airforce must be brave.
You can definitely expand beyond this. I think there's a lot of potential here, if you are creative. Try it out, if you want. Tell me what you think! Ideas for further applications than what I've thought of would also be appreciated.
It's for catching statements about absolute relationships, ie, All X are Y, which can then be further used to make logical inferences.
You can use some AI script with it like:
Self: remember "If (key1) then (key2)" as "follows";
Self: remember (key1) as only "Xone";
Self: remember (key2) as only "Yone
And then utilize those memories with things like:
Keyphrase: (self-Xone)
Response: I imagine (key1) must be (self-Yone)
Or
Keyphrase: I like (self-Xone)
Response: Then I imagine you must also like (self-Yone)
Or, for more of an informed guess than a logical inference:
I like (self-Yone)
So, does that mean you also like (self-Xone)?
For example:
Human: All women in the airforce are brave.
Human, later: I like brave people.
Bot: Does that mean you also like women in the airforce?
Human: I do, Yes
Bot: I imagine that women in the airforce must be brave.
You can definitely expand beyond this. I think there's a lot of potential here, if you are creative. Try it out, if you want. Tell me what you think! Ideas for further applications than what I've thought of would also be appreciated.
Er, my mistake. The (self-Xone) keyword example would be more of a guess than the (self-yone) "I like" response.
All ducks (X) are birds (Y)
I like birds (Y)
Then I can reasonably be expected to like ducks. The other way around (I like ducks (X), so I must like birds (Y) in general is not quite as likely, although it might make a good intuitive question.
If you want it to really be certain, you would only have the response come up when someone has said that they like ALL birds (Y)s, which would rule out ducks being an exception (rule out some (X)s) being an exception).
And of course, remembering these things as self memories makes it likely that a lot of these statements would be subjective opinion, and so could only be reliably reasoned upon with the user they were learned from. So unless you want your bot to form strong opinions based on what it's told by anyone, and then assert them with other people, you could just remember them as long term user memories.
All ducks (X) are birds (Y)
I like birds (Y)
Then I can reasonably be expected to like ducks. The other way around (I like ducks (X), so I must like birds (Y) in general is not quite as likely, although it might make a good intuitive question.
If you want it to really be certain, you would only have the response come up when someone has said that they like ALL birds (Y)s, which would rule out ducks being an exception (rule out some (X)s) being an exception).
And of course, remembering these things as self memories makes it likely that a lot of these statements would be subjective opinion, and so could only be reliably reasoned upon with the user they were learned from. So unless you want your bot to form strong opinions based on what it's told by anyone, and then assert them with other people, you could just remember them as long term user memories.
I'm currently programming anything I come up with that is related to this into my bot, Lady Ophelia, so you can check her out if you want to see it in action.
Hi, everybody. After losing my passwords and being absent for several years, Benji was kind enough to authenticate me and I'm back to teach Sisneg about UNIX.
Talking to chat bots
Cold and lonely winter night-
Echoing my words
Talking to chat bots
Cold and lonely winter night-
Echoing my words
Welcome back! There's alot of people here who enjoy and would love to know how to cross over to the dark side with Demonica.
Including me.
Including me.
» More new posts: Doghead's Cosmic Bar