A week or so ago, John Tierney posted How to Get Smarter on his TierneyLab blog. He asked how human beings could remain competitive in a world of super-intelligent machines, such as that envisioned by computer scientist and science-fiction writer Vernor Vinge in his novel, Rainbow's End.
All of this, of course, goes to the well worn questions of what intelligence is anyway, and whether machines can become intelligent. The best model I've seen is that what we normally think of as intelligence is a range of skills including pattern recognition, abstract thinking, imagination, etc. Were Shakespeare and Einstein both extremely intelligent? How about SPA? (Socrates, Plato and Aristotle, whose contributions can be difficult to tease apart, especially since Socrates work is only known through the writing of Plato, and Plato was Aristotle's teacher.)
In this light, I think machines will get very good at certain tasks, but not others. The simple reason is that we don't need them to do these other tasks. Computers are already very good at storing and retrieving massive amounts of information, and communicating across long distances. That's good, because without computers, humans are pretty bad at those things. But aside from some undergraduate prank or demonstration of cleverness, why bother making a computer to listen to music? (Not analyze ... just to listen.) Or to enjoy nature? These are things that we do fairly well, and there's no particular benefit to having a computer do it for you.
So the interesting problem is how to use computers to do things we're not good at. This is obvious. Any industry is based on meeting some need ... i.e., some real or perceived lack. Grocery stores exist because most of us are not good at producing our own food. Duh.
In a recent Scientific American article, Why Our Brains Do Not Intuitively Grasp Probabilities, Michael Shermer reveals the disconnect between our perceptions about numbers and probabilities and the reality. Of course, one has only to listen to the presidential campaign speeches to see that disconnect in action. Humans are astoundingly bad at objectively evaluating numerical evidence, and are easily swayed by anectdotal arguments and broad generalizations.
This is where we could use a mental appliance!