I agree with much of this article, except that I regard intelligent machines as either a threat or an opportunity, depending on how we adapt our society to them.
I quibble with one aspect of this discussion: the notion that computers will exceed human intelligence at some specific time in history, or perhaps over a decade or so in the middle of this century. I think intelligence is too complicated for that to happen. Let me be specific: In some ways, computers already exceed human intelligence. ENIAC, the first computer ever made, exceeded our ability to do arithmetic, which is surely a form of intelligence. Intelligence is not a single quality. Machines may gradually surpass us in some ways, while they remain behind in others. It might take decades or centuries for them to exceed us in most aspects of intelligence. Humans are the most intelligent species, but in some ways, for some purposes, other species are far more intelligent than we are. For example, if a dog could talk about what things smell like, she could tell us far more than the world's leading human expert on perfumes or cooking. I doubt that a machine could be designed that could track down a lost child by scent better than a dog can. A bird or a bat knows how to fly better any human pilot ever could. Humans have more "general intelligence" but for specific purposes, other animals have much better developed intelligence. Along the same lines, even if super-intelligent computers emerge I expect there will be specific areas in which humans can out-think them for a long time to come, possibly forever. I doubt, for example, that computers will ever understand human interactions such as politics or sexual attraction as well as we do. They may not be able to do pattern recognition well as we do. Especially face recognition. Although I have to say, Google's free photo utility program "Picasa" has built-in face recognition that I found uncanny. In a few cases it managed to identify faces more accurately than I did, although in most cases it did not. When I say people have more general intelligence than dogs or chimps, I mean we have more raw capacity. More brain cells. That does necessarily mean we have specific abilities or capabilities. This is similar to saying that the Watson computer has more general, factual knowledge of the world than any person, but people still have specific knowledge and the ability to make use of it and draw conclusions better than Watson does, in some ways. Okay, not enough to win at Jeopardy, but in some cases Watson made "foolish" errors that no human would make. I expect that in 30 years, intelligent computers will no longer make such errors. But we might still be better at sussing out some answers, just as dogs will probably remain better at tracking lost children. Whether these computers will be sentient or not is an entirely different question. Whether they should be deliberately designed to be sentient is both a practical question, and a moral one. - Jed
