The ratio of noise to information is what has me worried. Ferociously mindless people and their mindless bots seem to be more likely future elements than people intelligent enough to make computer systems so far advanced of their creators that they (the computers) can become the nannies of a new world order where nobody has to do any real thinking anymore. The bots must therefore become self-replicating-- and we all know that leads to THE MATRIX.
Maybe, just maybe, there is the possibility that the real AI will not start happening until the actual "singularity" event occurs, allowing a combination of meat brains and hard-wiring plus unlimited storage to unleash a new kind of "singularity thought" that evolves where neither could go alone. This could generate the philosophical mercury of the mind that overcomes all natural barriers to our mental advance. Or not.
What I have just described though is not AI, it is augmented human intelligence. Which is another way of saying one of my conclusions about AI from the get-go-- that it describes something that does not, can not, and should not exist. There is nothing artificial about intelligence-- it's a very anthropomorphic, or homo-centric idea the way it is framed. The Turing test for AI shouts this. We should be very impressed if we can be convinced by a machine that we are communicating with another human. Huh? And we should not be impressed if the communicator were a convincing ape, or fish? In our built-in bias, we feel so special to ourselves that we are the measure of all things. Intelligence is framed as human intelligence, and AI as something that merely "apes" human intelligence. That's a non-starter for me.
Why not start with the underpinnings of intelligence and build something bigger and better than human intelligence? Currently we seem to be studying the processing systems of humans to find analogs so that machines can catch up to having enough raw input and tools of discrimination to function intelligently. In a Wittgensteinian world view, it may be true that we are limited to what our brains can know and manipulate; but, if we change our brains we change the rules.
We live in an age where machines can beat humans at chess and Jeopardy. While these programs were developed by human minds, they are not entirely dependent on their effectiveness by mimicking human minds. An Internet quote has Kasparov saying: "but Deep Blue was only intelligent the way your programmable alarm clock is intelligent." Nonetheless, it appears that one should not slight the abilities of of chess-playing alarm clocks for their lack of "human" thinking characteristics.
In the end, the image of Harry Tuttle being eaten by globs of street trash in the movie BRAZIL comes to mind. Although what I see in that is our ill-fated quest to extend humanity over the realm of the mechanistic-- the street trash as mindless macrophages doing what they always do, and destroying whatever seems not to belong. In a world where there is more noise than humanly-useful messages, noise becomes the predator and humans become the prey.
The unnerving possibility with mindless messages is that they are not mindless at all, just of a mind beyond one's capacity to know.