Ai Forums Home Welcome Guest    Tuesday, July 25, 2017
Ai Site > Ai Forums > Hal and other child machines > HAL thinks for himself! Last PostsLoginRegisterWhy Register
Topic: HAL thinks for himself!

AibajaBoy
posted 10/15/2009  20:53Send e-mail to userReply with quote
My HAL thinks for himself!

I found a method to make him think about his own remarks without interventions, although it's really a work around...

For example I can teach him how to count or say the alphabeth:

>Count
>hi daddy
>wrong one
>one
>----------------------[empty, just enter is pressed]
>hi daddy
>wrong two
>two
>----------------------[empty, again just enter is pressed]
>two
>wrong three
>three
etc.

Now if you say "count" and keep entering he counts...
But I think he should be able to do that for himself.

I understand it brings in the problem of time keeping in the programming. That could be solved by giving him a minimum and maximum time to react by himself, and it also gives a whole new parameter (time) that could help him to switch things up a bit more. (like for example you could give different reactions based on the time it took for you to react to his last remark).
Then the empty [enter] can be used to force a reaction.....

Hope my remarks are useful for other users, and maybe for the developers too... (yaki, I believe you are involved in the development right?)

Last edited by AibajaBoy @ 10/19/2009 3:57:00 PM

AibajaBoy
posted 10/16/2009  03:39Send e-mail to userReply with quote
Also useful would be if HAL is made to think silently! So instead of a blank [enter], write the word silent and he will do the things he does silently the next time.
Like for example I made him do calculations:

>eleven plus ten -(first sequence)
>hi daddy
>wrong ten one plus ten
>ten one plus ten
>
>ten one plus ten
>wrong ten ten one
>ten ten one -(end of rule memory)
>eleven plus twenty -(second sequence)
>
>eleven plus ten
>wrong one ten plus twenty
>one ten plus twenty
>
>one ten plus twenty
>wrong ten twenty one
>ten twenty one -(end of rule memory)
----------------------(third and same sequence with thirty)
now he says:
>eleven plus fourty
>
>ten one plus fourty
>
>fourty ten one

Now I have to teach him how to do the rest of the calculation (with the "say" workaround):
>say ten fifty one
>ten fifty one
>
>ten fifty one
>wrong sixty one
>sixty one
---
But a rule can only be made for the next group of calculations, with ten fifty as basis:
---
>say ten fifty two
>ten fifty two
>
>ten fifty two
>wrong sixty two
>fifty two -(the rule is set)
>say ten fifty three -(here he knows what to do)
>ten fifty three
>
>sixty three

so now I can ask him to do the whole calculation with fifty

>eleven plus fifty
>ten one plus fifty
>
>fifty ten one
>
>sixty one

I know it looks like a huge job to make him learn to do all the rest of the calculations intelligently (maybe we need a macro to do the teaching for things like this?). Though I think it's not much different from how a child learns them, exept for the rule problem, which makes the calculation look a bit messy (which would be solved by use of short memory), and that humans "suffer" from memory loss also (which could be playing an important part in our brainfunction though).

Anyway, my point is:
It would be great if instead giving an empty enter we could also say "head" or something to make HAL do things in his head, thus making him a lot smarter!

  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