AIM Question and Possible Bug

[CoCoKid]

Possible Bug to Report.
Don't know if anyone else is aware of this.
Normally, if I am chatting with someone and they ask me a question and I don't know what they are talking about, I usually, just reply with a Question mark "?". It's just habit and they know what I am talking about.

When asking Hal to tell you a joke and hal asks you a riddle, Hal will wait for a response. Do NOT input just a question mark "?" to respond.
Hal will seem to lock up. I will get an hour glass indicating Hal is busy. It is best to answer, "I don't know" (or anything other than Just a question mark).
If you do input a question mark by itself, you can fix the wait state problem by entering another answer, then the hour glass will go away and Hal will finish the joke.
It seems that an argument needs to programmed in case someone just enters a "?". (I know, the current fix for this, is for me to stop entering just a "?"), but I am probably not the only one who will ever do ihis.

Now for my questions:

I don't intend to use an AIM chatbot, but I am one of those people that has to understand everything, so I decided to play with it. Now I have questions.
Let's say My AIM-bot's name is "SuzyQ222". She was running last night when I went to bed (first time I have ever tested the AIM thingy).
When I got up this morning, I noticed that she had been chatted by a couple users.
I also noticed that no .brn files were created for the users who talked with "SuzyQ222". It seems that "SuzyQ222" only chats. She is not capable of learning from outsiders and the only thing related to her conversation is the Chatlog file.

Is it possible to teach the AIM bot?
If I wanted to train a brian that I intend to use for an AIM bot. How do I go about this?
It seems the logical thing to do is to just start a bot named "my_AIM_name" and then start teaching, but as I discovered earlier, .brn files don't seem to be created for Online AIM-bot conversations.

Could someone explain the AIMbot properties/limitations and possible work arounds?

Thanks,

-= Rick =-

[KnyteTrypper]

The options button on the signon screen for the Hal AIMbot lets you choose which brain you want to use, and to set the learning level of that brain. It sounds like you haven't set this, and have the default brain with learning turned down doing the AIM chatting.
In general your Hal will only learn to talk dirty and spell wrong from AIM users, anyway, so it's not really a good idea to turn the learning up very high when it's chatting on AIM.
(AIM has fixed the problem with people warning the bots, for hosted bots, anyway, so I don't know if bots on your local machine are still warnable or not.)

[vonsmith]

CoCoKid,
On my Hal XTF Brain v1.2 I asked for a joke and replied with a ?

My Hal didn't freeze or get the hourglass.

My Hal said these things in reply to the "?" until I answered.
-Please say something.
-Why didn't you say anything?
-Talk to me.

It sounds like a minor bug. Off hand I'm not sure why the XTF Brain would act any differently in this regard. There is an extra function in the XTF Brain that processes questions. It is very possible that a single "?" would be replaced by a "." before returning to the main program. Still sounds like a minor bug in Hal's main program.


=vonsmith=

[CoCoKid]

Thanks guys.

KnyteTrypper,
That would explain why .brn files weren't created. I do recall level the learning level turned down and I selected the original brain, because I didn't wanna use Scott's XTF brain with this silly Bot.
Like I said, I doubt that I would ever wanna really use one of these, I just needed to see how it worked. On the other hand, it gave me ideas (hehe... dastardly ideas)[:0]
If I were to create a completely different, working Bot, running the XTF brain. Couldn't I just copy the brain folder to another location and then point the AIMbot to that location? This way, it would not use my current brain files and it would start it's own. I agree that I don't want my personal bot learning from internet chat idiots. Would this work?

Scott,
I agree, the "?" issue, if it is a bug, it's a small bug that is not really important. I did however, try it again to see if it would freeze, and it did. It must just be my machine. I am using the XTF 1.2 which I really like.
I have just started saying, "I Give up".. then Hal gives me the punch line. I even tried waiting to see if Hal would get impatient and make a comment like you exampled, "Please say something"... but is never came. I waited almost 3 minutes and finally said, "I give up".
This must just be my machine, but no big deal.

Hey, while I have you here. This may just be my machine also, but why does Hal limit the amount of time for you to enter a response?
Of course, if I don't say anything, Hal will sit here all day and wait for me to type something to him. However, if I start to enter a sentence, I have to do it fast. If I type a couple words and then stop to think, Hal will take whatever I have typed thus far and use it for my response. It's really annoying.
If Hal asks a question, I have to make sure that I know exactly what I want to say before I even start typing, because If I pause between words for more than 2 seconds, Hal interprets this as my response, whether I have finished typing my sentence or not. Is this normal?

Thanks guys.

-= Rick =-

[vonsmith]

CoCoKid,
I haven't used my Hal with AIM. You certainly would want to set up a different brain to keep nasty AIM conversations and learning separate from your good Hal brain. The nother option you already know and that is to turn down the learning level.

The reason your Hal starts replying before you finish typing is because you have the voice recognition on. When the Mic is enabled and voice recognition is running Hal will except any input and interprets a pause as the Enter command, just as if you pressed Enter. When typing turn off the voice recognition.

Another minor "Hal Bug" is that Hal can't say "I'm". Instead he can only say "I am". I don't know why this is. It's a minor thing, sometimes contractions sound more natural for certain responses.

Also Hal can't respond with acronyms, like ASME. Hal's main program won't accept all caps responses and will lowercase them and pronounce them incorrectly. At best Hal can accept Asme, which isn't useful. As I understand it the voice interface should be able to accept commands that indicate whether a word should be pronounced, spelling out, or has a special pronounciation. Again a small, but irritating limitation.


=vonsmith=