as many know here hal currently still makes false reactions to what it "thinks " is insults , when just as like happened to me recently with hal , i was speaking about something else and used the word crazy in it , and hal said to me "i am not crazy don't insult me !"
creating a a false positive .
could a fix be a more direct specific script of listed usage like " you are crazy" hal recognizes this as a "direct" insult (script wise ) where as not listed or anything else not specifically listed won't be generated as a insult . like are you crazy , who would do something like that. which is a more common sentence of expression rather then a straight out insult.
Anyway Maybe Robert can try redoing something this way within hal. i know he said he was going to work more on stopping the false positives of the math question reaction from hal.
I have talked with others on my thoughts of hal , and think i think personally hal needs the ability to do the same as what cortanna etc. does and that is be able to automatically access info. from the world wide web (with filters of course ) for answers , etc. i think once this can be accomplished hal ,can and will be a lot smarter and able to react better with users , while at the same time be an a.i. with hals own learned thoughts etc. which it can put forth conversation with the information it retrieves from the web . I also think this would create a major turning point with hal. and actually would in the long run create more sales of hal as an a.i.