That’s why I’m curious about what its other possible answers were. I’m trying to see what it might have picked up on to lead it to Toronto. I mean, is there anything you can see in there that would make Toronto a likely answer? The logic stumps me. It’s almost as if it was, indeed, giving a “joke” answer.
Both Toronto airports are named for people who served in World War I, not II.
One of them was a flying ace with 72 kills.
The main Toronto airport is named for Lester Pearson who was a prime minister of Canada.
edit - it turns out Bishop did serve in WW II as well as WW I.
OK, that makes a little more sense, then.
Still, it seems weird that the algorithm didn’t see a mismatch between the answer and the category title.
I wonder what percent Watson had for Toronto? And I wonder why they did not show the percents.
Not very high - the ??? after it indicated he wasn’t confident - probably wouldn’t have rung in if it had been in the regular rounds.
I thought that the programmers had attempted humor with Watson’s DD bets. When he lost the first one, I thought the individual digits resulting in his new (lower) amount were 3-4-5-6, although not sequential.
He came close the second time.
However, his “tranna” answer of FJ is just out there. Maybe a private attempt at humor at Trebek for being canadian, eh?
“He” was just on Conan talking trash about Ms. Richter, and Andy went off on “him”, smashing “him” to bits.
Cool–good to know. Hopefully there will be some actual wordplay categories in tomorrow’s game. It would be a shame if they did all that work for nothing!
Two questions. When they display the 3 answers, there’s a vertical line running through the percentage bars of each answer, somewhere in the middle. I’m guessing this indicates the confidence threshold required to ring in. But the position of that line changed depending on the question, so is it indeed a threshold, and if so, what was causing it to change?
Also, if he doesn’t have speech recognition and no human was feeding him data, how did he know when he had the correct answer and hence had to pick the next category? I guess maybe he could’ve picked a new category every time and they just edited that out.
It is the ‘buzz’ threshold. They never got into what makes it change. At least not when they were explaining how he worked on Jeopardy…the Nova special may have been different.
There is a human operator that signals him when to say the answer and when to pick a category.
This, I believe, is one of the major reasons Ken Jennings was able to have such a long run. Once he felt he knew an answer, he could ring in with a much higher success rate than his competitors. Not just faster reflexes, but over time a better rhythm for the pace of the game, a rhythm that his competitors could never get, since they were all brand new to Jeopardy.
I’m a little bit less unhappy with Watson’s clear buzzer advantage when I think how these champions used their buzzer skills to help themselves win millions of dollars.
Anyway, as much as it’s a challenge, it’s mostly a coming out party for Watson, to show how far computers have come in power and the ability to understand real language. It also highlights their deficiencies, when they get things wrong, it’s spectacularly wrong.
I wonder what people would say if Watson was losing? Probably something like “it just shows how far behind computers are compared to the human mind”
But what if Ken Jennings had a brain as big as a room?! Next year maybe.
When exactly did Watson get the answer in text form? Was it exactly when the speaker had finished saying the answer?
I unfortunately can’t find anywhere to watch the episode.
when the answer becomes visible for the humans to read it would also be sent to Watson.
You’re joking, right? That’s the entire point of this - to see if you can get a computer to understand a question. Saying, “well, once you understand the question, it’s simple!” is like saying, “if you can see the Moon, it’s trivial to fly there!”.
As for the “US” part of the category not being seen as important - how many times is there a category name that’s a pun or play on words or extremely misleading? It’s the question that is the most important, and that’s what programmers should (and did) spend their time on.
Dogging a machine that just destroyed the top two Jeopardy competitors of all time is perplexing.
I’m not joking, and I’m not dogging it either. I’m just puzzled that it couldn’t parse a very straightforward clue (no wordplay at all), and offered up an answer that clearly could not be correct as its “best” answer. I’m extremely impressed at how well it did in the rounds, under time pressure. You’re no doubt correct - it didn’t use the category as a limiting factor at all, even though it would have cut down the possible answers considerably.
Maybe after tonight they will explain the Toronto answer.
The announcer said that the other guy, not Ken Jennings, was undefeated. Is he still competing on the regular show or did he retire undefeated?