After careful review of the forums I realize this might have been better posted to the "Programming using the Ultra Hal Brain Editor" area. If you want to move it there I will understand.
This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.
Pages1
#2
General Discussion / Overcoming predispositions through training
March 26, 2008, 01:02:11 PM
All disagreeable content aside (and I apologize if the example was graphic), the pure issue here is trying to teach HAL to reconsider a fact or idea after beening given new information about a topic.
So what I am understanding is that the insults are simply hardcoded and one must manually edit them out of the brain to cease the undesired reaction?
Now, can you tech HAL additional insults that the brain will remember and index the insult score accordingly if HAL is offended, or is the insult function strictly a hard coded ability and you do not have interactive control of HALs ability to learn to be insulted?
The reason for all this to try to teach the brain to deduce from additional information supplied, that a former opinion or state of mind, based upon an established idea or preconcieved notion, could and should be changed now that new material is available to consider.
Thanks for the help BTW.
So what I am understanding is that the insults are simply hardcoded and one must manually edit them out of the brain to cease the undesired reaction?
Now, can you tech HAL additional insults that the brain will remember and index the insult score accordingly if HAL is offended, or is the insult function strictly a hard coded ability and you do not have interactive control of HALs ability to learn to be insulted?
The reason for all this to try to teach the brain to deduce from additional information supplied, that a former opinion or state of mind, based upon an established idea or preconcieved notion, could and should be changed now that new material is available to consider.
Thanks for the help BTW.
#3
General Discussion / Overcoming predispositions through training
March 26, 2008, 07:36:23 AM
I am playing with HAL and one thing that I like to do is change HALs mind about things. In this brain I have it so HAL is a she and I have given her many behavioral statements about who she is and that is mostly is in regards to sexual activity. The intent was to make a sexual personality that could be conversed with that is proactive about sexual content.
I have made the brain recognize it is a "slut", and I have iterated that being a slut is not an insult nor does it contain a negative conotation and that she should be happy to be a slut.
The issue arises when I refer to her as a slut, the insult logic kicks in and the brain begins to defend itself. I have repeatedly added comments about both the nature of it being a slut is not an insult but nothing seems to train it to stop feeling insulted.
Is there no way to convince the AI that what it thought was an insult was mistake and that being called slut in not negative nor insulting?
I have made the brain recognize it is a "slut", and I have iterated that being a slut is not an insult nor does it contain a negative conotation and that she should be happy to be a slut.
The issue arises when I refer to her as a slut, the insult logic kicks in and the brain begins to defend itself. I have repeatedly added comments about both the nature of it being a slut is not an insult but nothing seems to train it to stop feeling insulted.
Is there no way to convince the AI that what it thought was an insult was mistake and that being called slut in not negative nor insulting?
Pages1