http://www.typologycentral.com/forums/private-blogs/63752-universe-nonsense-2.html#post2129365
A couple of interesting questions to ask ourselves on the subject of robots: are we creating life? And are they in there present condition, alive? To the first question, I will answer with a resounding yes. In fact, we as humans, are effectively catalysts for the evolution of a set of entirely new lifeforms. To answer the second question requires a little more thought. The definition of life is not quite all that clear-cut as it is, but furthermore, I believe that to question whether something is alive or not, is not applicable for a simple "yes or no" answer. Rather than "yes, it is alive" or "no, it is not", there is much more ambiguity; in a sense, something can be half-alive, or be an entity that could be classified as having some traits of being life, yet not so much as for us to classify it as such in our ordinary paradigm.
Viruses come under this classification. It has traits of life, yet needs to be "awakened" by a host's cell. Merely, life as we know it is simply a distinction at a certain level (from inanimateness) - a level that we are used to seeing through familiar forms of life, such as mammals and amphibians. We may fail to recognize anything below this level, or perhaps even above - which may be a reason for the lack of encounters with extraterrestrials. So it is not the case of animate or inanimate, but varying degrees and shades of grey. Through this new distinction, it could be seen that computers have a level of animates similar to that of the virus; slightly below that of "familiar life". This correspondence emerges with a similar behavior of needing a host to perform its programming; rather, we are to the computer that the cell is to the virus.
To my mind, this human-causal evolution will continue, and computers will breed to such a degree that their programming will be able to match that in our own brains. This means building upon programming such that the computer is coded to intake millions of pieces of data and factors from the environment at once and constantly, in the same way we do. At present, the programming is too simplistic to take on traits what we typically consider a life-form. Fundamentally, the programming in our computers and those in our own brains are no different; just that they have emerged into existence from a different way. Only at a sufficiently advanced level of programming will robots behave exactly as what we view life to be. They will probably be completely autonomous, able to reason, perceive the world around them and themselves in relation to it, and most important of all - question reality.
This of course raises questions. Would these robots - who then would be our equals - be morally "good" or "evil", and what is the possibility of them turning against us? All of this depends wholly on the environment in relation to its evolution. Homo sapiens are its environment, so robots will develop around us, and how we program them to be. We have the control to encode a human morality, or perhaps they will remain utterly neutral. In my opinion - just as there are different subspecies of species due to genetics - there will be variants. Some will be good, some evil, some neutral. As we are its environment, we are the deciders of the creatures' (and yes, at this point we should consider them as such) behaviors. No doubt our own nature will lead to "morally evil" robots that have the set of data to behave destructively to its environment for ensured survival. This works in the same manner that some animals are naturally aggressive to other rivals, or territorial, or skittish. If this is the case, then yes, robots will turn on us, but not exclusively - just as part of their nature to be aggressive, and as such will kill lifeforms other than us too.
The second question is, how will they reproduce? Life by definition needs to reproduce in some way or it will obviously die out. At this stage of their advancement, a few things might occur. The first is that the robots might - either by deliberate programming, or an emerging, accidental property - be naturally predisposed to be co-dependent to humans. This means that they will be continually peaceful towards us (and as such, rule out the previous mention of "evil" robots), and that we will still be part of their role for survival; building them/repairing them. A symbiotic relationship may occur, with us enabling persistence of their existence, and them offering us protection and/or knowledge, among other, unknown things. Another possibility is that they will completely split off from humans, and build themselves, or re-use unbroken parts of dead ones.
Thirdly, a complex mechanism might be present at a future point that allows some sort of reproduction in the familiar sense of the word, although I cannot say how exactly this would function. I also believe that they will still be susceptible to death, as is the case with all life-forms, and are not exempt from loss of quality of their physical bodies and deterioration. It may, though, be easier for them to heal and repair themselves as a result of the nature of metal rather than flesh.