Ai Forums Home Welcome Guest    Thursday, November 23, 2017
Ai Site > Ai Forums > Hal and other child machines > New Idea Last PostsLoginRegisterWhy Register
Topic: New Idea

Grinktis
posted 7/14/2010  08:11Send e-mail to userReply with quote
I have an idea for a better way to teach Hal how to learn. Instead of just saying "wrong, (correct response)" you should just say wrong. Saying "wrong" would act like pain or displeasure and saying "good" would would act like pleasure.

If you were to say "wrong" when the response was wrong then he would try something else that he has learned from other experiences or from what you just said before.

If you were to say "good" or "right" then he would say that response more.

I say this because I feel like saying "wrong, (correct response)" only makes him a repeating machine, just repeating a line in response to certain words. I feel like this is not really working, no real intelligence will come from this current system.

Its like reading text is your only sense and whenever you do something wrong someone reprograms your brain for you by texting a word and then saying "this is what you should have said you big dolt and now say that from now on whenever I do this"


toborman
posted 7/14/2010  22:12Send e-mail to userReply with quote
 
Grinktis wrote @ 7/14/2010 8:11:00 AM:
I have an idea for a better way to teach Hal how to learn. Instead of just saying "wrong, (correct response)" you should just say wrong. Saying "wrong" would act like pain or displeasure and saying "good" would would act like pleasure.

If you were to say "wrong" when the response was wrong then he would try something else that he has learned from other experiences or from what you just said before.

If you were to say "good" or "right" then he would say that response more.

I say this because I feel like saying "wrong, (correct response)" only makes him a repeating machine, just repeating a line in response to certain words. I feel like this is not really working, no real intelligence will come from this current system.

Its like reading text is your only sense and whenever you do something wrong someone reprograms your brain for you by texting a word and then saying "this is what you should have said you big dolt and now say that from now on whenever I do this"

 
chatbot programs like Hal, based on the Eliza design, may become "knowledgeable", but will never become intelligent. this is not a flaw, it is a choice to make an entertaining program with a simple design.

 human mind map

tkorrovi
posted 7/15/2010  11:58Send e-mail to userReply with quote
 
toborman wrote @ 7/14/2010 10:12:00 PM:
it is a choice to make an entertaining program with a simple design.

 
Yes, and if it even was a True AI system, most people would not be able to do anything with it. It takes a long time until it would be possible to interact with any such system by simple chat. So it is apodictic that any AI entertaining program today can only be made with restricted capabilities.

 Artificial Consciousness ADS-AC project

pankajzone
posted 1/14/2012  16:54Reply with quote
Thanks for the info Grinktis. The post is very informative to me.

 SEO Company
  1  
'Send Send email to user    Reply with quote Reply with quote    Edit message Edit message

Forums Home    The Artificial Intelligence Forum    Alan and other chatbots    Language Mind and Consciousness  
Contact Us Terms of Use