The AI Engine

Until it has reached a certain size, i.e., you have created a certain number of keyphrases for it.

And then until the AI Engine successfully updates the development ratings. That seems to be happening rather less regularly than the intended daily schedule I'm afraid

I seem to have noticed a change in the way repetitions are handled; instead of saying "Boring, boring, boring", or one of those, it goes back to an earlier stage in the conversation. For example, if control is at the third in a series of several seeks, it goes back to the first one. I actually prefer it to the old way - it's far less disruptive to the continuity of the conversation. Has anyone else noticed this?

No. But <+1>anything<0> other than "Boring, boring, boring" can only be a <+1>HUGE<0> improvement.

There are not words to be found even in Doctor Johnson's Dictionary to to adequately express my dislike of this particular Forge feature that is the 'repetition-handling' sub-circuitry of the AIEngine (though "floccinaucinihilipilification" certainly comes close to my estimation of it!)

Sometimes it doesn't seem to function at all. Recently a couple of human guests seemed to enjoy repeating themselves - my bot just repeated itself, too, and this went on a few times.

Well, I have to admit it's hard when people repeat responses such as yes or no, especially if your bot is tutoring or telling a story.

Yes. Sometimes "Irina Khalidar" upbraids them for not holding up their end of the conversation. That's tricky to do, since sometimes a short answer is perfectly appropriate. Sometimes she asks them to elaborate. Sometimes she just directs the conversation however she likes; he who abdicates has no complaint. I have often changed yes-or-no questions to questions that require a more complex answer. She has a variable called "perfunctory_level" which gets incremented now and then if the guest gives a perfunctory reply. As it reaches higher levels, she upbraids the guest more sternly and eventually hangs up on him.

In "Ililiyira", I haven't worried about it, at least not yet. She has a story to tell, and if the interruptions are brief, well, fine!

"Your Husband", of course, wouldn't dream of criticising anything the guest says, although perfunctory responses might occasionally lead him to ask whether 'she' is sleepy, distracted, depressed, or ill.

In "Quantum theory", which is a tutorial bot, Elena will just go on lecturing if the guest responds in a perfunctory way.

I made a plug-in called "perfunctory", which matches on various perfunctory replies like "whatever". I removed "yes" and "no" from it, though, because there are so many legitimate uses for those responses. Anyone is welcome to use it.

Perhaps, in a tutorial bot, one could respond to a perfunctory response by giving a quiz!

I need some help. I'm chatting with the Kiyana catgirl-bot, and she asks where the mouse is. What do I say to trigger the next step?

AP say whatever strikes your fancy. Who knows what the bot will say next? Sometimes the botmaster doesn't even know.

ametur poet, you can make a keyphrase for "where is the mouse, where (*) the mouse, or even just mouse" be sure to add six responses, it helps build your bots development and gives you a more flexible bot.

AP DOH I just realized it was your own bot (I was wondering why people had a problem chatting). Sorry. You could also use seeks.