Jewish World Review Feb. 24, 2011 / 20 Adar I, 5771
Will the next Watson need us?
By A. Barton Hinkle
http://www.JewishWorldReview.com | Did Watson prove the Unabomber right? Is Skynet — the self-aware artificial intelligence from the "Terminator" movies — about to arrive? Sci-fi enthusiasts who grew up reading Omni magazine probably can't help asking such questions after IBM's supercomputer beat the two best "Jeopardy!" players in history at their own game last week.
This isn't the first time something like that has happened. Machines have bested humans ever since John Henry eked out a victory against a steam-powered hammer and then keeled over dead. In 1997, IBM's Deep Blue beat chess champion Garry Kasparov in a six-game match (Kasparov accused the computer of cheating). A decade later a chess program called Deep Fritz, running on a laptop, beat world champ Vladimir Kramnik without breaking a sweat.
Some have dismissed this as nothing to write home about, since chess is "just math." By contrast, Watson's performance required it to understand the nuances of natural language, which computers — being extremely literal-minded — tend not to do well. Or at least didn't used to do well. For the most part, Watson did.
So has Watson demonstrated true artificial intelligence? In one sense, not even close. IBM scientists worked extremely hard to refine a very specialized application — a program that can answer trivia questions. Watson may be able to tell you which burg is nicknamed Sin City, but it can't tell you if your purse matches your shoes, why Shakespeare is a better writer than Tom Clancy, or whether you should ask someone out on a second date. It's incapable of noticing that the job market for its skill set is extremely narrow, and deciding to take some classes in sales and marketing in case its position gets outsourced to a call center in Mumbai. At best, Watson is an idiot savant.
On the other hand, Watson does demonstrate characteristics indicative of genuine intelligence. It can learn. Not only that, it can identify what it needs to learn about. Even more telling: The team that created Watson could not understand everything Watson did. IBM's lead Watson researcher, David Ferrucci, has said Watson "absolutely [surprises me]. People say: 'Why did it get that one wrong?' I don't know. 'Why did it get that one right?' I don't know."
This, say those who think about artificial intelligence, indicates emergent complexity — a process by which simple systems give rise to much more complex ones — and it has led to a debate about whether Watson represents a sort of singularity.
The singularity is the point at which technological progress becomes so rapid human beings no longer can keep up. In one common formulation, humans create machines smarter than humans themselves. Those machines then create machines smarter still, and so on — the result being the rise of artificial intelligence that rapidly evolves far beyond human comprehension.
Perhaps the most notorious person to reflect along these lines is Ted Kaczynski, also known as the Unabomber. In his manifesto, Kaczynski writes: "First let us postulate that the computer scientists succeed in developing intelligent machines that can do all things better than human beings can do them. In that case presumably all work will be done by vast, highly organized systems of machines and no human effort will be necessary. Either of two cases might occur. The machines might be permitted to make all of their own decisions without human oversight, or else human control over the machines might be retained." He sees both possibilities as equally grim — the first because humanity would be at the mercy of machines, and the second because humanity would be at the mercy of a tiny elite in control of the machines.
Kaczynski is a lunatic. But he raises questions that have troubled widely respected non-lunatics, among them futurist Ray Kurzweil and Sun Microsystems founder Bill Joy, who, in an article for Wired on "Why the Future Doesn't Need Us," quoted Kurzweil quoting the passage above. Kurzweil himself takes a much more rosy view — speculating that technological acceleration eventually will make human beings immortal rather than obsolete.
Which one of them is right? We might find out sooner rather than later. Computing power is growing at an exponential rate. Human intelligence is not. Writing the other day in The Wall Street Journal, Kurzweil predicted that within a decade standard PCs will be able to perform the 80 trillion operations per second that Watson's 90 servers can. By then, artificial intelligence programs may be to Watson what the Xbox 360 is to Pong. At that point, those wondering whether artificial intelligence and human intelligence will merge or diverge may be able to find out by asking the machine on their desk.
Every weekday JewishWorldReview.com publishes what many in in the media and Washington consider "must-reading". Sign up for the daily JWR update. It's free. Just click here.
A. Barton Hinkle is Deputy Editor of the Editorial Pages at Richmond Times-Dispatch Comment by clicking here.
© 2010, A. Barton Hinkle