This week a machine captured our attention as it battled for supremacy against two human competitors. Watson, the IBM computer, prevailed over Jeopardy's finest, Ken Jennings and Brad Rutter. Jennings was quick to explain that "there's no shame in losing to silicon," he wrote on Slate. Watson and its similar prototypes will move on from here to big projects like improving health care delivery and smartphone technology. Scientists and technicians clearly proved something this week about ingenuity, progress, and communication. What's the big takeaway from this three-day experiment?
See the differences between man and machine: It was Watson's "human attributes that make him so compelling," says Joanna Weiss in The Boston Globe. But ultimately it behaved how you'd expect: heartlessly. The experiment showed "just how hard it would be to mimic the complexity of people," despite how actively we try to "turn our computers into friends." As the other contestants surrounded the machine at game's end to add some levity, "Watson was unfazed. He didn't get the joke, even though he took it like a man."
This was so utterly predictable: "Any seasoned horse handicapper easily could have predicted the outcome of the race," says a St. Petersburg Times editorial. Worries about a "takeover of human society are premature." For the humans, "there is no shame in coming up short against one of the world's foremost creators of computer technology," IBM. It didn't prove itself smarter than man, either,"just quicker. "The big winner in this contest was science."
We can't handle losing: So many people are "downplaying the win," says "R.M." in The Economist, arguing that it's no big deal and that man is still superior. It's "as if the man-made machine was a threat to our own self-worth." This rejection of Watson's accomplishment conveys"antipathy directed at scientists, academics, and experts," and "part of this modern anti-intellectualism stems from an unwillingness to accept our own inferiority." What this really shows is "our refusal to admit that America may not be uniquely great."
Watson will be to office and skilled employment what farm machinery was to field work In 1911 not so log ago 70% of USA population were involved in agriculture. Today it is around 3% in farm work. Watson type computing will do the same.
Then what will we do? WE could redefine what it takes to eat and have a house and get medical care etc. But currently we are moving the other way with the Tea Party types trying to pretend it is 1850 rather than 2011.
People released from work is the purpose of technology, this should not be used as a means of punishing us but freeing us from bondage.
I really see no recognition of this. Watson can be a liberator or a terminator. That is up to people. It was people who programmed HAl 9000 improperly. We do not really want to hear Watson saying "I'm sorry Dave I am afraid I cannot do that".
HAL 9000 was programmed that the mission was all important. What is Watson's mission? It is not a tech thing, it is only and totally a human thing.
Even without voice-recognition, Watson can replace research assistants, especially in law. Of course legal researchers use databases already, but Watson is a database that everyone can use because, for the first time, you really can ask a computer a question in plain English.
I'm not a Luddite, but all this technology is displacing employment, and we're not doing anything about it.
I believe that when compared to the best players, the "speed advantage" is less significant than a naive understanding would imply.
On the topic of non-infallibility: I've been a software developer for my entire life (54 years old, programming since age of 16) and was *transfixed* by the entire performance. I have no skin in the game and nothing but curiosity about what 15 Tbytes of RAM, 2880 cores and a big pile of disk can do. For me, at least, there wasn't a disheartening moment. Nothing but amazement, flubs and all.In fact, the nature of the flubs is a wonderful window into some of the logic of the program they were demonstrating.
Side note: Pressing the trigger prematurely is almost equivalent to red-lighting in a drag race: Premature triggering results in a delay of some time, possibly as much as a quarter-second, before allowing that player to record a trigger event.
More Coffee...
R/ PRONESE
The carbon footprint of this dinosaur from an aging technology company is profound while the mere mortals are rapidly sinking in the sinkhole of jeopardy.
The design of this dinosaur is more profound as it sequentially threads thru faking parallelized thinking of the mere mortals.
The system is built from 2880 Power/7 processors, so it is seriously parallel. It is not classical IBM big iron (which is no dinosaur, BTW... If you want 99.999% uptime, the Z series and its recent derivatives are the way to go).
The IBM researchers who took on this grand challenge make no claims of the system "thinking". Their claim is that the system is competitive with the best-known human players in a game well known to be challenging to humans.
Unless the 100-plus year history of consistent exponential growth of information processing power stops in the next year or two, the capabilities and carbon footprint of today's 2880-CPU Power/7 system is likely to be replaced in 10-15 years by a desktop system whose carbon footprint matches that of a 150-watt lightbulb, probably far less. Even today, multiple-teraflop computing systems can be purchased over the internet (search for "Nvidia Tesla") for five or ten grand. Those systems are also massively parellel running anywhere between 480-960 cores, each operating at a few billion operations/second.
Any TESLA would outperform such a dinosaur. Obviously, NVDA has better things to do than this watsonian gasp.
The OSes (AIX and z/OS) on these watsonian hickups are proprietary while real supercomputers are open with Linux. Heck, one can't even buy the hardware without the OS software. That is bundling and monopolistic by definition. Which on earth do they (IBM) price the same System z at twice the price in India relative to the rest of the world. Why? for market share and proprietary lock-in.