Posted on 02/16/2011 7:15:29 AM PST by SeekAndFind
Mankind put up a fight in the first round, but the second round was all machine. The good news: For once, Time magazine is right. The Singularity is near, my friends.
The bad news: We all know how this story ends.
The computer brained its human competition in Game 1 of the Man vs. Machine competition on “Jeopardy!”
On the 30-question game board, veteran “Jeopardy!” champs Ken Jennings and Brad Rutter managed only five correct responses between them during the Double Jeopardy round that aired Tuesday. They ended the first game of the two-game face-off with paltry earnings of $2,400 and $5,400, respectively.
Watson, their IBM supercomputer nemesis, emerged from the Final Jeopardy round with $35,734.
If James Cameron ever gets bored with making “noble savage” movies about Smurfs, he should do a Terminator prequel about the evolution of Skynet. First it wins game shows, then it moves into, er, social engineering, then it comes up with Stuxnet and takes down the world’s industrial infrastructure overnight. From there, it’s only a developmental hop, skip, and jump from burly killer robots with inexplicable Austrian accents emerging from the rubble.
Two clips for you, one from yesterday’s show of “Watson” fielding sports questions and the second a longer clip about how it works its magic. Fun fact: This isn’t the first time a man’s been defeated by a machine at Jeopardy. Word on the street is that Wolf Blitzer’s so bad, he once lost to a 2-XL.
The final jeopardy question last nite was what U.S. city....
And HAL said Toronto.
Watson should have had to actually *listen* to the question, parse it, figure it out, *then* press the button and answer.
The way it was run it got the question in a digital instant while the humans had to wait while Alex read it in a few linear seconds. No matter how you slice it there is a bias towards the computer in that setup which allowed Watson to ring in more quickly than the humans.
They didn’t have a chance, even though Watson got a lot of answers wrong.
In the end it was a pointless exercise. Recognition and retrieval is what computers do and the individual(s) who prepped the machine all but admitted that the only variable would be the timing of the buzz-in.
The Jeopardy! puns are a) usually lame and b) dead giveaways so claiming that the computer would need some sort of powerful A.I. to decipher them was spurious at best.
Jeopardy obtains all its questions from one encyclopedia.
How is this impressive when the computer just simply acts as a high speed reader?
I'm there.
Not the computer. The program that drives the computer.
Dumb...
Time to execute Spork Weasel!!!!
As a one-day Jeopardy champion, I have some insight on this issue.
For me and a good many other contestants, the most challenging part of the contest is not knowing the right answer. It’s ringing in at the right time.
You have to buzz in after Alex talks speaking and be the first one to do so after the little light up behind him goes off. If you buzz in before the light goes off, your button is locked out for half a second, which is an eternity under the circumstances.
Obviously a computer is always going to beat human reflexes at this game. I don’t know how they arranged the “buzz in” for Watson, but it has at least the potential to give the computer a huge advantage that has nothing at all to do with knowing the right answer.
They didnt have a chance..
Exactly, that is like a calculator versus man with paper and pencil to do math. No contest.
Great story line - but here's the problem. Computers don't have the human desire to dominate. You have doubts? Ask the computers who won Jeopardy if they 'enjoyed' the game. You won't even get a blink.
Make that: “You have to buzz in after Alex finishes speaking...”
Actually, if they want to be “fair,” they should have a human do the buzzing in for Watson.
And this whole concept is interesting why?
“Obviously a computer is always going to beat human reflexes at this game. I dont know how they arranged the buzz in for Watson, but it has at least the potential to give the computer a huge advantage that has nothing at all to do with knowing the right answer.”
They should have set it up so the computer received the text question the instant the host stopped speaking. I have no idea how it actually worked, but that would have evened things up a lot.
The other thing, which apparently isn’t quite “there yet” would be to have the “computer” (actually thousands of CPUs) process the host’s voice. That would be even more fair.
Actually I’m surprised they didn’t try voice recognition, the current systems are quite good.
NOVA Smartest Machine on Earth
When Watson was trying out for the show the host kept making fun of his responses when he got it wrong and the programmers took it very personally. "Why does he have to ridicule Watson?"
So, Watson has access to Google while in play.
That’s cheating.
“Actually Im surprised they didnt try voice recognition, the current systems are quite good.”
I thought this was what was happening, the whole point of the exercise.
“So, Watson has access to Google while in play.”
Nope, no Internet access.
Disclaimer: Opinions posted on Free Republic are those of the individual posters and do not necessarily represent the opinion of Free Republic or its management. All materials posted herein are protected by copyright law and the exemption for fair use of copyrighted works.