Jeopardy, Watson the A.I., and Maserati

[Yesspaz]

Hey, everyone.

So IMB Research has designed a computer to answer questions like a human, and it's taking on the two most celebrated champions in Jeopardy history this week. VERY interesting stuff.

Also cool, during the section when they're explaining Watson, the music is "Inventions" by Maserati, available at your local AM library.

[OverHillandDale(Admin)]

I've been watching this. I'd say it's advantage is not the database knowledge, but the ability to ring in faster.

it's kinda fun, though.

[julieval]

I am a big fan of Jeopardy and have been looking forward to this exhibition. I agree with Dale; being able to ring in faster is the key. Ken Jennings often would ring in and then take some time to come up with the answer during his winning streak. I find it interesting that WATSON totally took a dive on the Final Jeopardy question last night. Given the category of US Cities, it should never have answered Toronto. It would know Toronto was a Canadian city.

[dinosaur]

There is a Toronto, Kansas (pop. 312)

[OverHillandDale(Admin)]

Originally Posted by julieval

I find it interesting that WATSON totally took a dive on the Final Jeopardy question last night. Given the category of US Cities, it should never have answered Toronto. It would know Toronto was a Canadian city.

Are you trying to say that it's rigged! No....

But then he could have been fooled by the fact that they said "america" as opposed to the US, because technically the US & Canada are North America.

(This piece of useless information has been provided by Capt. Semantics)

[julieval]

Neither the category nor the clue mentioned America. The category was US Cities. I just found this:
David Ferrucci, the manager of the Watson project at IBM Research, explained during a viewing of the show on Monday morning that several of things probably confused Watson. First, the category names on Jeopardy! are tricky. The answers often do not exactly fit the category. Watson, in his training phase, learned that categories only weakly suggest the kind of answer that is expected, and, therefore, the machine downgrades their significance. The way the language was parsed provided an advantage for the humans and a disadvantage for Watson, as well. "What US city" wasn't in the question. If it had been, Watson would have given US cities much more weight as it searched for the answer. Adding to the confusion for Watson, there are cities named Toronto in the United States and the Toronto in Canada has an American League baseball team. It probably picked up those facts from the written material it has digested. Also, the machine didn't find much evidence to connect either city's airport to World War II. (Chicago was a very close second on Watson's list of possible answers.) So this is just one of those situations that's a snap for a reasonably knowledgeable human but a true brain teaser for the machine.

[julieval]

Originally Posted by dinosaur

There is a Toronto, Kansas (pop. 312)

How many airports does it have?

[tobytanzer]

Toronto's Island airport has recently been renamed Billy Bishop Airport, after a WWI flying ace. He's the dude who shot down the red baron...(it was NOT Snoopy)
Lester B. Pearson Airport is named after our 14th prime minister, who also served in WWI in the Royal Flying Corps, and was in the diplomatic corps in London during WWII.

FWIW - perhaps more trivia than you wanted to know.... but hey, aren't we Jeopardy fans all fans of useless trivia?

[Yesspaz]

Okay, if you want a good laugh, watch this clip. Conan did a skit about Watson last night.

http://video.teamcoco.com/video/index.jsp?oid=244025

[jtmckinley]

My $0.02: The important thing with IBM's Watson is the natural language processing and then the analytic software behind that question understanding and to be able give an answer and weight the probability that it is the correct answer, all within a couple of seconds so as not to "push the button" if the possible answer has a low probability of being correct. There's also some analytic processing in deciding how much to bet on the answer. In any event, I think this is a HUGE achievement in natural language processing, analytic processing and parallel processing. The specs on this beast are pretty amazing but even more amazing is the software IMHO, especially the parallelism of it.

[OverHillandDale(Admin)]

You are absolutely correct. I mis-remembered it. Saw that when I watched Wednesday's conclusion.
Stupid brain cells, who needs them anyway.

Oh, yeah. I do.

One thing I do like about this is that answering questions isn't so cut and dry. I guess we're only as good as the information that's being fed to us. Hey, that would explain a lot in this world, wouldn't it?

Originally Posted by julieval

Neither the category nor the clue mentioned America. The category was US Cities. I just found this:
David Ferrucci, the manager of the Watson project at IBM Research, explained during a viewing of the show on Monday morning that several of things probably confused Watson. First, the category names on Jeopardy! are tricky. The answers often do not exactly fit the category. Watson, in his training phase, learned that categories only weakly suggest the kind of answer that is expected, and, therefore, the machine downgrades their significance. The way the language was parsed provided an advantage for the humans and a disadvantage for Watson, as well. "What US city" wasn't in the question. If it had been, Watson would have given US cities much more weight as it searched for the answer. Adding to the confusion for Watson, there are cities named Toronto in the United States and the Toronto in Canada has an American League baseball team. It probably picked up those facts from the written material it has digested. Also, the machine didn't find much evidence to connect either city's airport to World War II. (Chicago was a very close second on Watson's list of possible answers.) So this is just one of those situations that's a snap for a reasonably knowledgeable human but a true brain teaser for the machine.

[TheFish]

How about stop watching TV, rise! Up off thy bloated arse! Go take in the good outdoors.

Sounds good to me.

[OverHillandDale(Admin)]

Originally Posted by TheFish

How about stop watching TV, rise! Up off thy bloated arse! Go take in the good outdoors.

Sounds good to me.

Yeah, sounds great. You go on ahead and make camp. We'll be right behind you.

[PeterG]

"I, for one, welcome our new computer Overlords."

[TheFish]

Originally Posted by PeterG

"I, for one, welcome our new computer Overlords."

That is a truly scary statement.

[spewie]

well it will make it better for doctors once Watson gets out of Medical school, if they mis-diagnose you they can just blame the computer.

I wonder if you can sue a computer, I am sure we will soon find out

[thegooze]

Yeah shure Watson may be faster than Humans, But can he bring you a cold beer in a tall glass?

[Yesspaz]

Now that it's all over, here's two great articles on the whole thing.

http://www.slate.com/id/2284721/pagenum/all/#p2

http://today.msnbc.msn.com/id/416325...ion/?GT1=43001

[gr8sho92]

As I watched the shows, and before I read articles about this event, two immediate things popped into my head and one writer had them both. The first was remembering that famous American folklore about John Henry. And the second was a more than passing resemblance to Arthur C Clarke's HAL and SAL 9000 computers.

And while it might be interesting to debate whether or not the game was fair (remember that Watson was fitted with a mechanical plunger as well), that wasn't really the point, but rather how something artificial could understand and interact in normal conversation while being able to process volumes of information and present mostly correct responses.

Talk about your PDA!!!

[VAXman(Admin)]

Originally Posted by Yesspaz

Okay, if you want a good laugh, watch this clip. Conan did a skit about Watson last night.

http://video.teamcoco.com/video/index.jsp?oid=244025

It's about time that somebody took a baseball bat to IBM. Long overdue and well deserved it was.