it just makes easily for a sometimes good plot. nothing more about it.
however were not any closer to real, self aware and acting, AI now than we were 20 years ago.
talking shit about "singularity" being right over the corner is stuff best left for hipster cafes. we only have fairly simplistic choice machines available now and there's no clear path to self aware AI and far less a route to build an evil scheming plotting AI.
and if one actually READS THE FUCKING ASIMOV BOOKS about self aware robots and such is that they're mostly about how people will not like to be lead by a machine even if the machine is GOOD for them in every possible way(the other ones are just detective stories that use the robot acting rules as plot devices).
it would be much more for the day to worry about scenario described in a short story "A logic named Joe". basically, in that a helper artificial intelligence system gets it's limits removed, so you can ask it for instructions to do anything from how to kill your neighbor without getting caught to how to get some person to have sex with you or how to steal billions of money without getting caught. why? because that story has a plausible AI. the AI itself doesn't have any motive in that story, it just tells people answers to their questions. http://en.wikipedia.org/wiki/A...
Joe is very tomorrow and today short story, truly visionary. I highly recommend reading it since it's pretty short and a staple in scifi history. it's very today since people are asking google already that kind of questions - and it was published in 1946 - and most importantly the AI isn't some genie like in many other stories of robots and AI from the era.
I'm not saying real AI is unattainable, with motives and all that. but I find it much more plausible that we have some bioengineered neo humans before that - and most probably, some AI's which look after the other AI's plans.