Hero wrote:
All four scenarios of M superiority of H, M-H merging, H-M reliance, and H dominance of M, are all much much more likely than M simply killing off H.
The first general AI will certainly be in a very naive state when it is turned on, since you can't get real sentient behavior out of the systems we have today, which act kind of like ants or bees: they don't need a lot of training, relatively speaking, but are very good at only a few things.
It will need a teacher to impart behavior, and that teacher will most likely be the US military. As such, it will probably be trained to have some inclination towards wiping out things it doesn't like outright and surviving at all costs, as well as a good working knowledge of deceit. You wouldn't expect anything other than highly calculating psychopathy, right?
(I'm not 'afraid' of this scenario btw)
_________________
"You can take me, but you cannot take my bunghole! For I have no bunghole! I am the Great Cornholio!"