• Welcome! The TrekBBS is the number one place to chat about Star Trek with like-minded fans.
    If you are not already a member then please register an account and join in the discussion!

Psyched About Watson?

thestrangequark

Admiral
Admiral
Is anyone else looking forward to the upcoming Jeopardy tournament that will be featuring the IBM-built AI machine named Watson? The science geek in me (the one who is soon going back to school for a Neuroscience phd) has been absolutely giddy with enthusiasm for weeks.

On Nova Science Now tonight was a segment about the development of Watson, and it's very cool.
 
Oh . . . I was about to say this belongs in the TV and Media forum. Thought you meant Emma Watson.

She’s cute.
 
Is anyone else looking forward to the upcoming Jeopardy tournament that will be featuring the IBM-built AI machine named Watson?

I heard about this some time ago, but I don't know what Jeopardy is! Is it a general knowledge quiz?

If so, I think Watson would do well to point it's big blue antennae towards trekbbs, and tap into the wealth of information the posters have imbued this board with over the years. :)
 
^My god! Jeopardy is such an institution of television game shows here that it positively blows my mind to here that someone hasn't heard of it!

Jeopardy is a trivia show (the best EVER!), yes, but will provide a great challenge for Watson. The primary setback to artificial intelligence in many ways has been human language -- it is so subtle and complex that computers simply haven't been able to figure it out. Jeopardy is famously difficult, with a broad range of topics. What will make it an especial challenge for Watson, though, is the use of language employed in the questions. There are puns and slang, contestants must come up with portmanteaux on the spot, and players must phrase their answers in the form of a question. Here is Ken Jennings, one of the two greatest Jeopardy champions who is going up against Watson, in his first game:
[yt]http://www.youtube.com/watch?v=9qipHjByR58&feature=related[/yt]
 
Watson's a Question Answering AI. What this means is that it's designed for answering questions using an algorithm to determine what the best answer is. The biggest handicap Watson will have is in having to answer quickly since it will have to comb for answers.
 
Not necessarily. Watson has a massive amount of prossessing space at its disposal, which allows it to easily determine a number of possible answers in under two seconds -- the amount of time it takes a jeopardy contestant to buzz in. Time isn't really a problem for Watson, it's the ambiguity of human language that poses the greatest challenge.

After amassing a number of possible answers Watson uses statistical analysis to choose which one is the most likely to be correct. While this strategy is effective, it does nothing to aid Watson in the understanding of the quirks of language, meaning that while he buzzes in quickly enough, he sometimes gives obsurdly wrong answers -- for example, answering the question, "This was the first non-dairy creamer," with," What is milk."
 
Apparently we read different things. I read an article awhile back that emphasized that reaction time will be a problem due to the database sifting in order to come to the right answer, that human language is only one of the factors making it difficult for a computer like Watson to find the right answer. So, if you can imagine, because of the human language and the way we link things, a computer like Watson doesn't have that natural ability, so what they did to help with that was use multiple algorythmns to narrow down and corroborate answers.


Err, you know, nevermind. I think we're both looking at this the same way, only we're using different words lol



I think this is the article I read:

http://www.nytimes.com/2010/06/20/magazine/20Computer-t.html?_r=1

Quote from the article:
it must do more than what search engines like Google and Bing do, which is merely point to a document where you might find the answer. It has to pluck out the correct answer itself. Technologists have long regarded this sort of artificial intelligence as a holy grail, because it would allow machines to converse more naturally with people, letting us ask questions instead of typing keywords. Software firms and university scientists have produced question-answering systems for years, but these have mostly been limited to simply phrased questions. Nobody ever tackled “Jeopardy!” because experts assumed that even for the latest artificial intelligence, the game was simply too hard: the clues are too puzzling and allusive, and the breadth of trivia is too wide.
I figure the questions that will be the hardest will be the ones with ambiguous answers, with some words meaning different things in different contexts.
 
Last edited:
^Hahahaha, we are looking at it the same way! It's a wonder computers can understand as much language as they do, given that we supposed masters still have difficulty understanding eachother, even when we're saying the same thing!

Thanks for the article! :)
 
Apparently we read different things. I read an article awhile back that emphasized that reaction time will be a problem due to the database sifting in order to come to the right answer, that human language is only one of the factors making it difficult for a computer like Watson to find the right answer. So, if you can imagine, because of the human language and the way we link things, a computer like Watson doesn't have that natural ability, so what they did to help with that was use multiple algorythmns to narrow down and corroborate answers.


Err, you know, nevermind. I think we're both looking at this the same way, only we're using different words lol



I think this is the article I read:

http://www.nytimes.com/2010/06/20/magazine/20Computer-t.html?_r=1

Quote from the article:
it must do more than what search engines like Google and Bing do, which is merely point to a document where you might find the answer. It has to pluck out the correct answer itself. Technologists have long regarded this sort of artificial intelligence as a holy grail, because it would allow machines to converse more naturally with people, letting us ask questions instead of typing keywords. Software firms and university scientists have produced question-answering systems for years, but these have mostly been limited to simply phrased questions. Nobody ever tackled “Jeopardy!” because experts assumed that even for the latest artificial intelligence, the game was simply too hard: the clues are too puzzling and allusive, and the breadth of trivia is too wide.
I figure the questions that will be the hardest will be the ones with ambiguous answers, with some words meaning different things in different contexts.

^Hahahaha, we are looking at it the same way! It's a wonder computers can understand as much language as they do, given that we supposed masters still have difficulty understanding eachother, even when we're saying the same thing!

Thanks for the article! :)
All I can say is thank goodness it wasn't named "Connery." ;)
 
Haha yeah, exactly. I wonder if they'll have a question with the answer of 42, if Watson will get it :D That would likely confuse it considering it's referring to a number, except that we're not really looking for a number. What fascinates me is what the computer does in order to get the right answer, like this for instance:

Ferrucci showed me how Watson handled this sample “Jeopardy!” clue: “He was presidentially pardoned on Sept. 8, 1974.” In the first pass, the algorithms came up with “Nixon.” To evaluate whether “Nixon” was the best response, Watson performed a clever trick: it inserted the answer into the original phrase — “Nixon was presidentially pardoned on Sept. 8, 1974” — and then ran it as a new search, to see if it also produced results that supported “Nixon” as the right answer. (It did. The new search returned the result “Ford pardoned Nixon on Sept. 8, 1974,” a phrasing so similar to the original clue that it helped make “Nixon” the top-ranked solution.)

And if you refer to page 5, you'll see where I got the idea that Watson would be slowed down.
 
I heard about this some time ago, but I don't know what Jeopardy is! Is it a general knowledge quiz?
My god! Jeopardy is such an institution of television game shows here that it positively blows my mind to hear that someone hasn't heard of it!
This Weird Al Yankovic video will give you an idea of how Jeopardy looked back when the show had its cheesy old set and cheesy old Art Fleming.

[yt]http://www.youtube.com/watch?v=BvUZijEuNDQ&nofeather=True[/yt]
 
Well, the first episode with Watson was pretty neat, but he kinda fell apart in the last half or so. It's too bad he wasn't fitted with speech recognition, might have avoided that little oops on that one decade question. :lol:
 
If you are not already a member then please register an account and join in the discussion!

Sign up / Register


Back
Top