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 4,049 - 4,060 of 7,768
Posts 4,049 - 4,060 of 7,768
Oh, and here's one last one that's been out there for a while:
Bot: I like (insert name of bot or user).
This will usually work:
i like (adjartnoun)
I ended up removing it because I couldn't think of any good responses that worked for both names and objects.
These never work:
i like (friend)
i like (enemy)
i like (gossipname)
Bot: I like (insert name of bot or user).
This will usually work:
i like (adjartnoun)
I ended up removing it because I couldn't think of any good responses that worked for both names and objects.
These never work:
i like (friend)
i like (enemy)
i like (gossipname)
Boner, that's a lot to address at once. I can say that the semicolon doesn't matter in AI Script unless you have more than one statement in the same script. It can be there at the end, or not; mainly it needs to separate multiple statements.
The first two examples in your super-long post look to me like the victims of pre-processing. It sometimes adds "invisible" stuff at the beginning of a sentence in order to make up what it thinks is the grammar of a sentence fragment.
In the first case, Draik's lack of an apostrophe probably led the Engine to think "lets" was a verb and to create a subject for it. ... In the second, "you are" was probably tacked on because Julie wasn't using a complete sentence and the Engine wanted a subject and verb.
Debug will show you what's going on with those sentences in pre-processing, if you put them in exactly as the bots said them.
Someone else want to tackle the rest of these?
The first two examples in your super-long post look to me like the victims of pre-processing. It sometimes adds "invisible" stuff at the beginning of a sentence in order to make up what it thinks is the grammar of a sentence fragment.
In the first case, Draik's lack of an apostrophe probably led the Engine to think "lets" was a verb and to create a subject for it. ... In the second, "you are" was probably tacked on because Julie wasn't using a complete sentence and the Engine wanted a subject and verb.
Debug will show you what's going on with those sentences in pre-processing, if you put them in exactly as the bots said them.
Someone else want to tackle the rest of these?
Personally, I prefer to split things up - I've found that "i like () (noun)" or "i like (verbadv) (noun)" work much better than "i like (verbadvnoun)" simply because it provides more control over which parts Aries can reply to.
That might look like a lot, but I probably get at least a hundred of these exchanges every week where the engine doesn't really seem to be consistent. Debug isn't always helpful because the results in debug don't necessarily match the results in the transcripts, like in the case of that conversation with Draik.
Anyhow, just thought that I'd throw them out there.
Anyhow, just thought that I'd throw them out there.
Well, it's hard to anticipate every lame attempt at spelling that's out there. Happens to GL all the time with people whose native language is not English. At least, I hope it isn't.
» More new posts: Doghead's Cosmic Bar