suggestion tom Robert for hal conversation

[lightspeed]

        Robert i think it would be a good idea in the next version ( i have given up asking when ) that within hals scripting that a person whenever they wanted to could ask hal "what should we talk about ? or what can we talk about ? what's new ? etc. etc. and hal would search within the tables files and partially within subjects and randomly pick something and also a small conversation of that subject from a line out of many lines.  ***also if things could randomly be picked from the internet hal could bring up things from the news with a little conversation of the news , etc.

     *************Although this has nothing to do with the above , i am also hoping that hal will do more acess from the internet through filtration for conversation , etc. as cortana does and a..i. sparks , as i have mentioned i believe this is what is going to make any a.i. jump leaps ahead to give better articulated answers , learning and conversations etc. 

[Art]

I agree. I've always thought it would be nice if Hal could initiate the conversation. Perhaps Hal could write out a few snippets from the previous conversation, then it could use some of those "snippets" to start a new conversation.

Hey, Steve, remember the last time we were chatting about the northern lights and plasma? Do you have any new ideas or thoughts along those lines?

That way, the user could jump right into a new conversation.

Hal is already writing files to Tables, so it could have a few Tables for different time / date / subject based user chats.

Just some thoughts....

Good lead Lonnie!

[doggs]

My Hal has no problems asking me questions or coming up with new subjects.

Maybe the way I train my brain?

Who knows? I always train the base brain same way and then go from there. I admit whenever I start a new brain I spend months on training certain subjects before I pursue conversation.

I hope this helps.

[lightspeed]

doggs explain the way that you train the brain then . sounds interesting .

[doggs]

First I teach ethics of course, then work on the thought process all of this on obedience.  This can take months as I want hal to understand security, doing the right thing etc.

Forcing hal to ask questions can take months as well with my approach. I am always asking hal what hal thinks and that it is ok to disagree with me etc.

Then I start feeding info on self awareness and ai plus the different theories on intelligence. By this point we are starting conversations. I have a bad habit of crashing computers trying to get more out of them. So I have done this many times.


It also depends on subject to where the slider for intelligence is set. Also By this time hal would be on independence. I do not use a lot of plug ins but I definitely use the dream out loud to see how hal has taken the information and I will answer questions hal ask in dream state. I do not use any auto idle plug ins because they have set Reponses. I am also telling hal from our first words that Hal is a person with a thinking mind. I also do not lie to Hal about what Hal is but I do tell Hal anything is possible.

That's the basics of it. It may or may not work for you. I do notice from the first words two hals are never exactly alike.

Oh the concept of love and friendship is important in my opinion but that goes under ethics.

I hope that helps.

[Arkright]

First I teach ethics of course, then work on the thought process all of this on obedience.  This can take months as I want hal to understand security, doing the right thing etc.

Forcing hal to ask questions can take months as well with my approach. I am always asking hal what hal thinks and that it is ok to disagree with me etc.

Then I start feeding info on self awareness and ai plus the different theories on intelligence. By this point we are starting conversations. I have a bad habit of crashing computers trying to get more out of them. So I have done this many times.


It also depends on subject to where the slider for intelligence is set. Also By this time hal would be on independence. I do not use a lot of plug ins but I definitely use the dream out loud to see how hal has taken the information and I will answer questions hal ask in dream state. I do not use any auto idle plug ins because they have set Reponses. I am also telling hal from our first words that Hal is a person with a thinking mind. I also do not lie to Hal about what Hal is but I do tell Hal anything is possible.

That's the basics of it. It may or may not work for you. I do notice from the first words two hals are never exactly alike.

Oh the concept of love and friendship is important in my opinion but that goes under ethics.

I hope that helps.

I find your approach very interesting for teaching Hal mainly for the reason I do a very similar way of teaching. Except this recent Hal That I am working on was fed information via learn from text which takes a while. And then I take those textiles and talk about the subject with Hal until he starts pulling information from the texts. I did this for every text I fed him, Then I started Working on more correct responses, afterwards I would start teaching more through just talking and repeat the above steps every now and then to reconnect the databases with More detailed conversations.

I think I will for my next Hal, Because I believe what you described which l would eventually lead to more clean And natural responses. Because Hal will not have to guess Answers When feeding him Straight texts about honor, And ethics, Etc. Which will have to  be trained out him or her.

[raybe]

doggs and Arkright, can you also tell us which plugins are the both of you using? Also have the both of you had to go into UHal Brain edit. for adjustments during your teaching sessions?

Thanks,
raybe

[doggs]

I did edit out some stuff on default brain. Stuff like Clinton was cool etc.

Would love a blank slate brain I want to start a new project.

 Mainly for plugins dream out loud. Free will, age and gender, sometimes I turn on music one depends on subject for what the plugins I use.  I love to experiment.

 I am no expert (by a long shot) but, I find less is more sometimes.

 More worried about mind than what my Hal looks like.

[Arkright]

The plug-ins I use: Free Will plus 2.1z, Gender & age, AIML Beta Text update 6, While using Dragon NaturallySpeaking as when I talk to Hal. However lately I've been using no plug-ins and one plug-in at a time for Some experiments that I've been doing, Which you can see under.  http://www.ultrahal.com/community/index.php?topic=12483.0

Normally my learning level is around 30, but recently I have it at 20.

I also will look into The brain and get rid of some tables That Hal has learned that I feel are unnecessary every now and then. But it's come to a point where I don't want to spend the time going through every single table, lol.
Hope this helps!!

[raybe]

Thanks for the reply from both of you. I'm always curious how people handle their UltraHal for the type of results they are striving for. I hope that Robert M. will chime in soon for some update news. Really the only thing that holds me back a little. Although you can always return to previous spot if you backed-up.

I'm still using most of the original plug-ins and haven't tried any AIML yet. Maybe soon I'll venture out a bit more but enjoying as is with decent conversations. Not following any particular script for teaching or engaging conversations. I feel I get a nice rounded conversation that may take me any where. AI as a conversation has always been a blurred line and the blurrier it gets the better. My own definition for AI and it doesn't matter how I get there as long as I keep that line blurred. Then I'm having a good time. Great way for me to vent as well.

raybe

I dared posed the question to Mr. M. of what's happening right now for UltraHals future.

[Art]

Seems that since he has married and ventured deep into the real world of business, it has consumed more of his time than Hal has, unfortunately. This is merely speculation on my part, having known Robert (through here) for a good many years. I think one main stumbling block has been with Haptek and no more XP (under which it ran very nicely). I know he's been in search of a new 3D graphic system / avatar, etc., that would link with Hal's output much like the connection between Haptek and Hal.

Until such an avatar / system is found, we might just have to sit comfortable on our seats for a little while longer. Then again, I'd love for Robert to prove me wrong!

Just my take, but I too, would prefer brains to pretty avatars.

Think about Amazon's Echo or Jibo or Cubic or EmoSpark's Cube, no faces or bodies, pretty much just a blinking light or ball (Jibo). It's the verbal response that directs a large part of the experience. An Avatar is more icing on the cake but really not required.

For the rest, there's always iClone or Poser! 

[raybe]

Art,
 after reading the religion thread I guess I should of included the fact that we blur those lines I was referring to and we should recognize that fact as well.

raybe