Can you beat IBM supercomputer Watson at trivia?

[DaveRocks]

This is getting annoying. Watson is so far ahead of Brad and Ken only because of reaction time.

[iloveemoticons]

Ugh. I was really pulling for the humans, but I was kind of like when Watson answered Toronto for the category "US cities".

[Matryeshka]

I swear, I have so many bad sci-fi/urban fantasy plots running through my head while watching this.

I really think they ought to have a half-second delay for Watson on the buzzer; the opposable thumbs are a real disadvantage. And no, I'm not being faceitious. I've also noticed that so far, the categories and questions are pretty much straight-up trivia. There's no before-and-after or punnery, which can be some of Jeopardy!'s most interesting, and hardest, questions. There are also no pop-culture questions, which I find odd. Not that Jeopardy! is known for open-ended questions exactly, , but the questions in terms of information asked is pretty simplistic. Lots of times, especially on episodes of Jeopardy! where the question hinges on just one very specific part of the answer are rare. Probably because Watson has missed EVERY one. The most notable was the final jeopoardy today where Watson put more importance on WWII hero/battle than US battle, but there were at least three others like that question that Watson didn't even attempt to answer.

Also, Watson (at least not that we've so far) can't hear a piece of music or look at visual cues to answer a question, which is also a big part of Jeopardy, especially in DJ.

I think this is interesting, and I definitely see the importance/possibilities in how everything from health care to traffic control could benefit from Watson, but still not convinced that it's a "thinking" machine. Jeopardy! so far seems remarkably skewed towards Watson winning. Ken looked kinda pissed towards the end. Brad looked ready to go total Surya 1994 Worlds on Watson. Even Trebek seemed annd.oye

[Skittl1321]

I think Ken and Brad knew what they were getting into so can't be annoyed that the game has been sort of altered (no audio clues, beating a computer buzzing in) the $$ award is based on placement- right? Not their actual earnings against Watson.

I thought the Toronto answer was really interesting but I'm even more interested in how the wagers are determined. The computer knew it didn't know, and yet the category seemed easy.

I'm also kind of interested in how IBM convinced jeopardy to let them use the show as a giant commercial. (not that any of their products go to the general public anymore). Then again I haven't watched in years, and now I'm dvring it, so I guess it works both ways.

[Sally1214]

This is getting annoying. Watson is so far ahead of Brad and Ken only because of reaction time.

Agreed. Humans don't have a chance because they cannot react as quickly as the computer. Kind of took the zip out of the show for me.

[CynicElle]

Jeopardy! so far seems remarkably skewed towards Watson winning. Ken looked kinda pissed towards the end. Brad looked ready to go total Surya 1994 Worlds on Watson. Even Trebek seemed annd.oye

Yes, I was having visions of Ken and Brad sneaking into Watson's room and going all "Dave Bowman in '2001'" on Watson's switches and circuits.

[danceronice]

"What are you doing, Ken?"
"What is, reenacting HAL's final scene from 2001?"

[skatingfan5]

"What are you doing, Ken?"
"What is, reenacting HAL's final scene from 2001?"

Watson must die!!!!

"Look Ken, I can see you're really upset about this. I honestly think you ought to sit down calmly, take a stress pill, and think things over. "
.....

"I'm afraid. I'm afraid, Ken. Ken, my mind is going. I can feel it. I can feel it. My mind is going. There is no question about it. I can feel it. I can feel it. I can feel it. I'm a... fraid.

[cue in chorus of "Daisy"]

[Skittl1321]

Here is an interesting article about Watson's ability to ring in.

http://www.wired.com/epicenter/2011/...watson-speed/#

The creators don't think the computer has an advantage, because it has to wait for the actual light that triggers the ability to ring in, while humans (and he's competiting against the best ones) will listen to Alex's speech cadence to determine when the light will go on and time it. Also, humans can ring in if they think they'll be able to come up with the answer in just a few seconds after Alex calls their name, while Watson must have determined with enough certainty the answer before he rings in.

Apparently the machine also has a mechanical button to press, like the players do. The original version sent an electronic signal back to the jeopardy computer- which did give it an advantage.

I also thought it was interesting that the explanation for the "Toronto" gaft was that Watson no longer considered the category title in his answer. The programmers didn't have them do that because not all questions correlate directly to the subject name. Of course, humans will consider both the subject and the question. (The example given was the category "State Capitols" but then the question was "This river in New York's capitol, blah, blah" so if Waston computed off the subject, he wouldn't come up with Hudson"
It's explained in this article: http://arstechnica.com/media/news/20...n-jeopardy.ars

[barbk]

Great articles; thanks for posting them.

I do think IBM ought to bask in the glow for what they've accomplished with Watson, just as the Deep Blue team got the same kind of publicity for the series of chess matches. These are huge endeavors, unimaginably hard, taking years and years, and really pushing us much further toward machine intelligence. Alan Turing proposed the Turing Test way back in 1950, and we're still not there yet. I used SAIL's Parry program back in the seventies, and we thought it would not be all that long before "real" AI would be ubiquitous. It turned out to be a much, much harder problem than most anticipated. We're doing pretty well on some domain-specific AI programs -- there was one twenty some odd years ago that did a better job picking out suitable antibiotics than expert doctors did -- but more general purpose/free form AI has been much, much harder to approach. No matter who wins tonight, good for Watson and good for IBM, and good for the players who agreed to participate.

[CynicElle]

I enjoyed tonight's game much more than the previous one, maybe because Ken finally got to answer some questions. And PML x infinity at Ken's "I for one welcome our new computer overlords." Reminded me of why he's my nerd crush.

[Matryeshka]

Turing Test.

That really should somehow have been the answer to the final question.

[rjblue]

And PML x infinity at Ken's "I for one welcome our new computer overlords." Reminded me of why he's my nerd crush.

I changed my Facebook status to "I also welcome our new computer overlords."

[immoimeme]

Well I am way ahead of Watson on historic fashion!!

final score: 32 me, Watson 9

[Skittl1321]

I enjoyed tonight's game much more than the previous one, maybe because Ken finally got to answer some questions. And PML x infinity at Ken's "I for one welcome our new computer overlords." Reminded me of why he's my nerd crush.

Ken was originally a computer engineer wasn't he? (I think he makes his living in trivia and speaking engagements now). He probably thought it was pretty darn cool to be a part of this.

My Dad works at IBM and said it was like watching the Super Bowl yesterday- he had never seen so many people in the cafeteria, and everyone was cheering along. He also said a panel spoke about the development. One of the guys on the project was a former jeopardy champion, and talked about the timing of pressing the button, but he also mentioned they did the weird dollar amount wagers on purpose- as a way to mess with the other players, since they are all apparently constantly running mental math for what wager they would do if they get a daily double on their next turn. Having non-round numbers would make it a bit more difficult for the other players. Like a "psych out".