AI has traditionally been represented in Hollywood as an existential threat to humanity. Predictably, the counter-trend chose to depict them as homo humane – moral beings dedicated to humanity’s welfare, or at the very least, choosing not to interfere. In doing so however, we are merely projecting human psychological quirks upon their minds; “minds” that are fundamentally non-human.
To be sure, nobody really knows how consciousness arises (or even how to define it, strictly speaking). Increasing neural complexity along the evolutionary path, at one point, led to rudimentary self-awareness. Instead of a series of involuntary reflexes, the trait allowed the animal to take cognizance of his surroundings, and actively make use of its neural prowess to manipulate it in order to avoid danger and obtain food - the stepping stone towards intelligence. It reduced the turnover time that natural selection took in order to get the critical survival behaviour that had suddenly become indispensable. What would have taken generations to program into the animal for it to mindlessly execute, could now be partially improvised by the animal itself, making consciousness a chief candidate for optimization through natural selection. Origins aside, it is clear that our modern consciousness is an emergent phenomenon, the result of millions of years of evolutionary tweaking.
Similarly, our psychological impulses too have been shaped by innumerable, and mostly unaccountable evolutionary factors. Given that we can't possibly hope to recreate all the factors that led to our current psychological state, there is absolutely no reason to assume that an AI which becomes self-aware somehow, will share any of our expectations of behaviour. Why, for example, would it want to create something, follow orders, or have even an iota of curiosity? Why would it even want to survive? A consciousness brought about by artificial means with possibly no survival instincts, cannot be “obviously” expected to have any desire.
All of this assumes that the self-awareness we are used to is a distinct something that can be reached through multiple paths, and not something bound to a very specific evolutionary process. Even if it were the case, will we even call something that doesn't want to survive, conscious? Doesn't our very idea of consciousness hinge upon our perception of free will? If that is an illusion, might not consciousness be one too? Are all our efforts then, directed at making machines delusional? And, how much time will a “true” AI take to realize this?
(Image courtesy of Victor Habbick at FreeDigitalPhotos.net)