Jeopardy: Ken Jennings vs Brad vs Watson (IBM Computer)

tejdog1 · **Posted:** Tue Feb 15, 2011 8:28 pm

Yesterday was somewhat fair.

Today Watson just destroyed everyone.

bygranddesign · **Joined:** Fri Sep 22, 2006 8:09 pm **Posts:** 5576

watson dominating team human. i wonder if that ridiculous answer of 'what is toronto' for final jeopardy in which the category was U.S cities was the first indication of machine becoming sentient .. maybe that was Watson's first attempt at humor. A virtual wink if you will.

SHPIDA · **Posted:** Wed Feb 16, 2011 2:43 pm

I LOVED this. Ken, Brad and a computer in one Jeopardy match. It seems the people had a better chance when you couldn't just "google" it. If there were two or three words you could "search" a database for, the humans have no chance. It was only the ones that you had to do some "thinking" because there were no actual "keywords", like the final Jeopardy, that the computer is out of its element.

rejoice1 · **Joined:** Tue Sep 19, 2006 5:33 pm **Posts:** 4744

My heart goes out to Ken Jennings. his facial expressions were that of sheer frustration. It's been compelling TV, to say the least

Azmetaholic · **Joined:** Sun Oct 01, 2006 10:03 pm **Posts:** 11569

I found this post on a message board and thought it was interesting

Quote:

The IBM Watson Jeopardy challenge is bogus. Granted the computer can be as smart, or even smarter than the Jeopardy champions. The dirty little secret that IBM or Alex Trebek does not tell you is that there is a physical element to the contest. When Watson knows an answer, it can trigger the push button instantaneously – perhaps with only a few microseconds (millionths of a second) delay. When a human knows the answer, pushing the button is not instantaneous – there is a delay, perhaps a quarter or half a second delay (milliseconds) – much slower than Watson’s electronic response.

If IBM was totally honest about this contest they would have measured human response times to an answer, then built this delay into Watson’s response.

As the contest now stands one of the things it proves is that an electronic response is faster than a physical response. Duh!

SHPIDA · **Posted:** Wed Feb 16, 2011 5:29 pm

Azmetaholic wrote:

I found this post on a message board and thought it was interesting

Quote:

The IBM Watson Jeopardy challenge is bogus. Granted the computer can be as smart, or even smarter than the Jeopardy champions. The dirty little secret that IBM or Alex Trebek does not tell you is that there is a physical element to the contest. When Watson knows an answer, it can trigger the push button instantaneously – perhaps with only a few microseconds (millionths of a second) delay. When a human knows the answer, pushing the button is not instantaneous – there is a delay, perhaps a quarter or half a second delay (milliseconds) – much slower than Watson’s electronic response.

If IBM was totally honest about this contest they would have measured human response times to an answer, then built this delay into Watson’s response.

As the contest now stands one of the things it proves is that an electronic response is faster than a physical response. Duh!

Is this confirmed to be true? In one of the videos it seems they built a buzzer thingie for Watson. I'm not so sure this guy is correct.

bygranddesign · **Joined:** Fri Sep 22, 2006 8:09 pm **Posts:** 5576

i think it is an amazing thing that IBM has done. Obviously this is not strong AI like from what we see in science fiction films and books ... but it is a beginning to that revolution. Some Jeopardy questions have a certain nuance to them that will confound Watson right now but if you believe in the power of exponential growth it’s not hard to believe that watson will eventually understand tone, nuance, sarcasm ..etc

TrueBlue15 · **Joined:** Thu Dec 10, 2009 9:06 pm **Posts:** 393

Azmetaholic wrote:

I found this post on a message board and thought it was interesting

Quote:

The IBM Watson Jeopardy challenge is bogus. Granted the computer can be as smart, or even smarter than the Jeopardy champions. The dirty little secret that IBM or Alex Trebek does not tell you is that there is a physical element to the contest. When Watson knows an answer, it can trigger the push button instantaneously – perhaps with only a few microseconds (millionths of a second) delay. When a human knows the answer, pushing the button is not instantaneous – there is a delay, perhaps a quarter or half a second delay (milliseconds) – much slower than Watson’s electronic response.

If IBM was totally honest about this contest they would have measured human response times to an answer, then built this delay into Watson’s response.

As the contest now stands one of the things it proves is that an electronic response is faster than a physical response. Duh!

This can definitely be true. If you just focus on Ken you will see that he is pushing the trigger button right away on most of the questions but Watson beats him to the answer. It seems he is getting frustrated with it.

tejdog1 · **Posted:** Wed Feb 16, 2011 7:41 pm

I wanna see Watson do some Before and After, and Crossword Clues.

Like this

this was the home of the brooklyn dodgers, directed by Phil Robinson, starring Kevin Costner and Ray Liotta

dalek · **Posted:** Wed Feb 16, 2011 10:40 pm

TrueBlue15 wrote:

Azmetaholic wrote:

I found this post on a message board and thought it was interesting

Quote:

The IBM Watson Jeopardy challenge is bogus. Granted the computer can be as smart, or even smarter than the Jeopardy champions. The dirty little secret that IBM or Alex Trebek does not tell you is that there is a physical element to the contest. When Watson knows an answer, it can trigger the push button instantaneously – perhaps with only a few microseconds (millionths of a second) delay. When a human knows the answer, pushing the button is not instantaneous – there is a delay, perhaps a quarter or half a second delay (milliseconds) – much slower than Watson’s electronic response.

If IBM was totally honest about this contest they would have measured human response times to an answer, then built this delay into Watson’s response.

As the contest now stands one of the things it proves is that an electronic response is faster than a physical response. Duh!

This can definitely be true. If you just focus on Ken you will see that he is pushing the trigger button right away on most of the questions but Watson beats him to the answer. It seems he is getting frustrated with it.

Then what about the times where one of the other guys answers a question incorrectly and then Watson buzzes in afterward?

Mangey · **Posted:** Wed Feb 16, 2011 11:43 pm

We've been having a discussion about the buzzer issue on the National Puzzler's League mailing list, and it just strikes me as besides the point. The smaller point is that Ken Jennings has a huge buzzer advantage as well - you have to not only react fast, but you have to time it. You are not allowed to buzz in until the question is read, and then for a very small delay afterwards. If you buzz in early, you are locked out for half a second, and someone else will surely buzz in. Being good at Jeopardy involves getting that timing down, and returning champions always have that advantage.

But the more important point is that actually figuring out the answers in a timely fashion is what is interesting, not beating Ken Jennings at Jeopardy. This task is incredibly difficult. Watson has to do a good bit of sophisticated natural language processing to figure out what the questions are even asking, and to work its way through metaphors and puns. Once it does so, it has to access an incredible amount of information, and sort through it to find the relevant facts (this is *way* harder than it sounds). Plus it has to do this all in a timely fashion. Early versions of Watson took upwards of 2 hours to answer questions.

Jeopardy, like chess playing AI programs before it, is a good game to test out a range of abilities. It is not an end in itself. I think the focus on the buzzer issue distracts from what is really interesting about the program because it focuses too much on Jeopardy. To say that it only shows we can build computers that are faster than humans is to ignore the fact that it is what Watson does in that small interval which is so impressive.

TrueBlue15 · **Joined:** Thu Dec 10, 2009 9:06 pm **Posts:** 393

dalek, I admit I don't much about computers. Obviously Ken or Brad were faster than Watson in the situation you brought up. Common sense would say that Watson took longer to process the questions that it did not answer right away, which gave Ken or Brad the opportunity to answer before Watson could process the whole thing and buzz in?

Jeopardy: Ken Jennings vs Brad vs Watson (IBM Computer)

NYFSers Currently Online