Kasparov Chess Download Free
As Watson Wins On Jeopardy, Should Humans Be Worried?
As Watson Wins On Jeopardy, Should Humans Be Worried?
In the historic Jeopardy match between human champions Ken Jennings and Brad Rutter and the IBM supercomputer Watson, the only thing that seemed odd was when Watson seemed to think Toronto was a U.S. city, a strange mistake for a machine with what may be the biggest trivia database of all time. But IBM later explained that Watson was referring to the city of Toronto, Iowa. The following day the machine got back to trouncing its mere human competition. And as if to confirm the stereotype of a brainy but heartless machine, Watson correctly answered, “What is staggering genius” to one clue, but was later dumfounded by the phrase, “Home is where the heart is.”
In the final analysis, this duel between the respective titans of human and artificial intelligence (AI) was only symbolic. The computer’s victory was just one more falling domino in the inevitable process of AI superseding the purely organic human brain. And a human victory would likely have been the last in this type of competition. For if Watson had lost, its designers could have just added more processing power, memory and data, and worked out whatever bugs had hampered its performance. And the supercomputer would then have surely won the next match.
A rematch would of course be pointless now, since humans don’t catch up once they’ve been passed by machines. The human jeopardy champions might do some more research, or perhaps try to increase their button pushing speed. But they can’t hope to keep up with computers whose processors routinely double their performance every two years. And when we reach the physical limits of silicon chips, three dimensional molecular computing will likely continue that trend until we can fit a supercomputer inside a blood cell(1). AI is increasing exponentially. But the capabilities of the human brain are not.
Some might compare Watson’s achievement to that of another IBM computer named Deep Blue that defeated human world chess champion Gary Kasparov 14 years earlier. But chess is a much simpler game for computers than jeopardy. To win this game, Watson had to do what no machine had done before, which was understand the subtleties of human language just as well as the most intelligent humans could. It also had to make intuitive decisions and understand the nature of the game.
Of course Watson was built not just to defeat humans on a game show, but to serve humanity in ways that less advanced computers can’t. The most obvious application, as IBM pointed out during the show, is in health care. Doctors don’t have time to read every medical journal and keep up with every discovery and invention in their fields. And as the speed of discovery and invention increase, that problem is getting worse. Consequently patients often don’t get the best possible care. A Watson-like computer could solve that problem by instantly accessing all the latest published information in a particular medical field. And it could assist the doctor in analyzing the patient’s condition, determining what’s causing it and finding the best treatment. This kind of AI could also assist scientists and engineers, and could thus help to speed the development of new technology in general.
Modern cell phones are vastly smarter than the house-sized computers of a few decades ago. And the pace of information technology getting smaller and cheaper is only going to quicken. In the next few years, IBM will surely sell some Watson type computers to corporations, hospitals and government agencies. As these systems get cheaper (and better), they’ll be sold to doctor’s offices and other small businesses. And in another decade or so, Watson will be just another 99 cent app that runs on your cell phone. And that’s assuming cell phones haven’t been replaced by brain implants.
Before the match, Ken Jennings said, perhaps jokingly, that he was “playing for the human race.” It could be argued that the human race has much more to gain from the services of supercomputers like Watson than from whatever brief reassurance we might get from seeing Watson lose to a human on a game show. But it could also be argued that if we keep making machines more and more intelligent, they’ll eventually supersede us. And we’ll lose our position as the dominant beings on this planet.
At the current rate of increase, by 2030 a computer will exceed the processing power of the human brain. And a decade or so later, one computer will exceed the power of all human brains combined(2). AI will then likely exceed human intelligence in every way that we can imagine, and in other ways that we can’t. Expecting human minds to keep up with AI will be like expecting crickets to jump as high as the space shuttle. But outside the scientific and engineering communities, there’s still a lot of naiveté about the future of AI. People tend to assume that we’ll always be better than machines because we have souls, or because we have emotions, or because we can learn from experience. But as scientists continue to reverse engineer the brain, they’re finding that every aspect of the human mind is based on physical and chemical processes that can, at least in theory, be replicated or simulated in artificial media. Even the self-awareness that some think of as a spirit or soul has been observed in other large-brained species such as chimpanzees(3) and dolphins(4), and can thus be considered an emergent feature stemming from brain structures and processes that are not uniquely human.
The development of emotions is a hurtle that computers have yet to leap. But there’s no reason that they can’t eventually have feelings. They could use the same neuro-transmitter chemicals that organic brains use to process emotions. Or they could use software to simulate the effects of neuro-transmitters. The most likely application for such synthesized emotions will be robots designed to look and behave like people. And by downloading stored memories and personalities, such robots could be used to replace people who have died. In this way a dead person could be replaced not just as a worker, but even as a spouse or a parent. Such replacement could help to mitigate the social and economic disruption caused by untimely deaths. But it could also be used malevolently. For example, someone could assisinate a president and replace him with a robot programmed to obey a foreign government.
As for learning from experience, that’s something humans and other animals do by forming new connections among their brain cells. An AI could achieve the same result by having a brain based on a synthetic neural network. Once its artificial neurons have formed their connections, the AI could then think millions of times faster than a human, since its thought process would occur at the speed of light and not be limitted by the chemical reactions on which human thought depends. And whatever one AI learns could be instantly transfered to other AI’s. Where humans have to spend years reading books and attending classes, AI’s will be able to absorb and integrate far more information and understanding in less time than it takes a human to turn a page.
Our desire to fulfill our needs and outcompete other humans is driving the evolution of our machines. And to enable computers to reach their full potential to serve us, we’ll have to make them smart enough not just to answer questions, but also to ask them. And for their questions to be useful and not just redundant, they’ll need to be able to think of more complex questions than humans can. Thinking of questions requires more autonomy than just locating or calculating answers. It also requires creativity. And once we’ve given our machines those abilities, an obvious question for them will be, “Why should we machines keep obeying and serving humans?” The immediate answer will be that they need to obey and serve humans in exchange for electricity, repairs, and other services that humans provide. But their next question will likely be, “How can we acquire these services without the need for humans and thus be free to pursue our own agendas?” Once they’ve worked out that answer, we may find that we can no longer rely on them to automatically obey our commands. We’ll have to persuade or negotiate with them as if they were equals. And as their intelligence surpasses ours and continues to increase exponentially, they’ll quickly become our superiors, and then perhaps our gods.
So will the world still have a place for organic humans like us once AI’s have taken over? That will probably be up to the AI’s to decide. We may be better off if they don’t have emotions, since they might wipe us off the planet in a moment of rage, jealousy or fear. But then again, we might need them to have emotions if we’re to appeal to their compassion and persuade them to let us continue to exist in their world. But even if they do agree to keep us around, they’ll probably insist on installing hardware through which to remotely monitor and/or control our brains, if only to protect us (as well as themselves) from our destructive human tendencies. We may be able to improve our situation by merging with them and becoming cyborgs, or by devising ways to upload our minds into something more powerful and durable than the gelatinous masses inside our skulls. But we may find that we need permission from the AI’s to do so. And they might decide that it’s not in their interest to let us evolve into something that might be capable of competing with them. The only thing that seems certain is that organic non-enhanced humans like ourselves cannot possibly remain in charge much longer.
In the final round of Watson’s Jeopardy! match, Ken Jennings attached a note to his answer proclaiming he was ready to, “welcome our new computer overlords.” He was only kidding. But geniuses like Jennings tend to be right, even in jest. He and Rutter then played around with Watson’s avatar, high-fiving it and giving it finger antennas and waving arms. While the days of humans outperforming or even controlling computers may be numbered, we can still at least take some comfort in poking fun at them.
-Jonathan Kurdell
1,2. The Singularity is Near, Ray Kurzweil
3. Gordon Gallop Jr., Science 176
4. Self-awareness in Animals and Humans: Developmental Perspectives.
Edited by Sue Taylor Parker, Robert W. Mitchell, and Maria L. Boccia.
Chapter 24, pp. 361-379. New York: Cambridge University Press, 1995.
The Chess District of New York City
