Ok maybe I'm missing the significance of this as all I can come up with is "publicity stunt with no purpose." For those who didn't see, Jeopardy put Ken Jennings and another multiple time winner (his name escapes me) up against a computer as the third competitor to see who truly is the Jeopardy master in a 2 day tournament. The two problems that directly came to mind were,

#1 Someone or something has had to program said computer with the knowledge to be able to answer the questions. If said computer has access to the internet, then it would have the knowledge of several billion minds, not just the one mind that the other two would be playing with.

#2 A computer these days is capable of doing trillions of calculations in seconds, so by that, who decided what the computers response time would be to push the button and answer the question?

I realize this was all in fun and the money raised was donated to a worthwhile charity, but again, what's the point of playing trivia vs. a computer? At least Kasparov vs. Deep Blue had the uniqueness of the singular human mind with it's creativity, cleverness, and the ability to think outside the box, vs. the computer's plan based on logic and probabilites.

#1 - the computer did not have access to the internet, that was stated upfront. It relied on the data in its memory.

#2 - that's what I thought was bogus about the whole thing. The computer was fed the question electronically as soon as the bell went off. Nanoseconds later it had the answer.

To be fair, they should have delayed the computer whatever decimal seconds it would take a human to read the question and squeeze the button, maybe some average response time determined by measurements from years of correct answers.

From a report I heard on NPR, the big thing about this was Jeopardy style questions and answers are a different type of logic then chess. Chess can be written as a mathematical problem of odds and probabilities. Jeopardy required the understanding of language and even double meanings and puns. The special tournament was done as a test of the computers language skills, not so much a real test against humans. That part was to get people interested.

It was a publicity stunt, but understanding human language is an extremely tough problem for computers. Considering the fact that Watson can do billions of operations per second, it is telling that the humans could still beat him from time to time.

That said, there was an interesting criticism from Noam Chomsky about Watson:

GS: As the world's leading linguist, what are your thoughts on Watson, the robot that will be appearing on "Jeopardy"? This appears to be the most advanced form of AI to date.
NC: I'm not impressed by a bigger steamroller.
GS: I assume that "a bigger steamroller" is a reference to Deep Blue. Watson understands spoken language and adapts its knowledge based on human interaction. What level of AI would be required to impress you?
NC: Watson understands nothing. It's a bigger steamroller. Actually, I work in AI, and a lot of what is done impresses me, but not these devices to sell computers.

Also, according to Stephen Wolfram of the Wolfram Alpha answer engine, Watson does not actually understand speech, he's just getting answers by brute force searching of text documents in his enormous database.



It was interesting to see how pathetically he failed Final Jeopardy in the first contest.

In the next century, our computer overlords will get quite a chuckle when they take a nanosecond to review comments like these.

I think the answer to the question is singularity.

http://www.time.com/time/...38,00.html


I think the answer to the question is singularity.

What is singularity?



[This message has been edited by Russell Bonine (2/17/2011)]

I watched the Monday episode. All the techno-speak aside, I found it to be boring because of Watson's monotone and lack of a real face. The ultimate bore-fest would be to watch 3 computers compete.

The long answer is long, so I won't get into it.

The short answer is that the Singularity is when Artificial Intelligence gets so good it starts designing its following iterations and will eventually get to a point where Humans, unable to comprehend what is being developed, will have no ability to predict the future or actions of the newly created AI's.

This is actually a much harder concept to wrap your head around then it seems.

Imagine explaining a simple blackberry device to someone living in 1650. You'd be describing impossible magic to them.

So just try and contemplate what sort of technologies or devices could come from a designer with 1000 times the capacity of the smartest group of the smartest humans to have ever lived.

It would appear as magic to you and I.

To be fair, they should have delayed the computer whatever decimal seconds it would take a human to read the question and squeeze the button, maybe some average response time determined by measurements from years of correct answers.

FWIW, no one gets to buzz in until Alex finishes reading the question and a light goes on behind them letting them know that it's okay to buzz in. If you buzz in early, it locks you out for like a second. Ken Jennings admits that a lot of the reason that he won so much was that he perfected buzzing in the instant the light came on. Since the contestants can read the clue much faster than Alex reads it, they generally know if they have an answer before it's time to buzz in. That said, no one's reflexes are faster than the computer after that light comes on (although there is some delay since Watson had to actually actuate a solenoid or something along those lines that mechanically pushes the button just like the humans do). I think that most of the questions that Watson did not buzz in first on were ones where he was still working on the answer or didn't have a good answer, but the humans had plenty of time to read the question.