There's a very interesting article that says that scientists have approached to programm robots with routines that are self-aware and can interact in complex and arbitrary environments using wireless technologies to cooperate with each others(as machines) and with other wireless networks.
Here's the link:

http://www.eurekalert.org/pub_releases/200...r-fpa111005.php

Enjoy:)

The term "self-aware" is a bit overused in this case. The technologies in question wouldn't be sentient, intelligent artificial beings, but would instead be programed to determine what something will colide if they don't take action, and thus would trigger a defensive response. It's nothing exceptionally new or groundbreaking, and there is the problem of them thus not being able to communicate with objects that don't have wireless technologies (for example, obstacles, animals, older cars, etc.).

even though we have come a long way in AI, i would say it will be impossible to find out when a actual break through will come where robots could be integrated in human society. IE (terminator, I, Robot movies)

It's being debated if it's even possible, but assuming it is, it's true that it could be dangerous, so AI would need certain morals in order to survive in human society, and then humans would have to learn not to look down on them (if that's even possible; we haven't stopped doing that to eachother yet). I'm guessing that the first AI won't be in a robot like in most movies, but will instead be a gigantic quantum supercomputer with robots remotely attatched to it.