Where to find a good A.I. 'person'

[Kedas]

I was checking this A.I. chatbox http://aimovie.warnerbros.com/

 

'It' is quite stupid although it says it isn't

Does someone know a better program ?

 

I wonder if they put emotions in it would it be based on a formula or just based on 'data tables'.

would you be able to tell the difference?

 

with 'data tables' I mean:

if X or Y then I'm sad.

if A or B or C then I'm happy.

with formula I mean:

all inputs are presents.

 

an other question: is it possible to feel without the ability of imagination?

[YT2095]

I`ve done a little work with Neural Cybernetics (but am NO expert).

for a start your program would need masses of memory, and it should be able to write to this memeory as it "sees fit" to update itself (a bit like neural nets the more hits the stronger the bond) like on Google )

secondly you`ll need something akin to possitive and negative reinforcement, Good result/Bad result from which it "judges" it`s activities. these are then writen to memeory and so it will slowly "evolve" and become more "inteligent".

now the progs I used to write used an Int RND(time)*1 backup when both possibilities were equal (machines "lockup" at this point without a "decider") if in the case of a Robot however, you could use external stimuli as the "decider" until it builds such a database as to "think" for itself, you`de be quite surprised however the amount of times your initial prog will reach an impase tho! (hence the need for something to swing the descision).

it`ll also need to be Goal orientated!

as for "feelings"? who can say?

 

hope that helps a little?

[Kedas]

isn't the reason why it get possibilities that are equally valid because the calculations to determine it are too simple?

 

so with more complex decisions it would keep runnng?

 

I will take care of the emotions in it, just keep it running

[YT2095]

Ok, for instance, one of my 1`st attempts at A.I was a simple cursor on a screen, it`s objective was to get from the top left of the screen to the bottom right.

the screen would generate a random maze however, making it more difficult obviously. and because it was random, occasionaly there was no way it could reach it`s goal, but each time it did a maze succesfull it would store it`s number sequence and patern problems and success failure at each juncture. Untill one day it did something TOTALY BIZARE! it reached an Impase, it couldn1`t reverse because that would A, be against it`s goal, and B,get it nowhere. so it jumped out of screen mem parameters, into the free mem of the computer itself and performed a "wrap round" !!!

a bit like how your cursor jumps back to the left of the screen on a new line. it was NEVER taught to do this or even how, infact it was against it`s original parameters set from the 1`st time i ran it. but it altered itself. that RND(time) decider seed kicked in and it "saw" that off screen it could still "survive" AND bring it closer to its goal. In effect, It Cheated! I was in hysterics laughing at it with my jaw part on the floor too, I couldn`t beleive it

and that was on a 6502 based BBC micro with only 32k ram