Yesterday started a three-day series of Jeopardy, but what's different is that we have two of the shows most prolific winners facing off against IBM's "Watson". Here is a practice match from last month:
Is anyone watching the Jeopardy "Man vs Machine" series?
I think it's utterly fascinating, and although I'm going to miss the broadcast tonight and tomorrow night, I'll be watching it through the video IBM says they're going to post on Thursday.
I appreciate how they're showing Watson's top answers and confidence, as it really helps to understand Watson's strengths and weaknesses throughout the categories. Also interesting is how Watson's employing a bit of a "shotgun" method of category selection, moving around the board horizontally rather than vertically. That seems to be a specific tactic intended to throw Rutter and Jennings off.
Should be interesting to see how Watson handles the Double Jeopardy! round. It's apparent it was already struggling a little with some of the higher-value answers in the Jeopardy! round.
Yes! I remember Trebek mentioning on Friday that they were doing something completely new, but I definitely did not expect this to be what they were doing. I'm definitely very intrigued and interested to see not only how the games play out over the next two nights, but where this technology goes.
" I enjoyed it. I hope today's episode has more questions then back story about it. "It's seems that they are contnuing from yesterday and moving on to double jeopardy. Although, I wonder if they are going to do final jeopardy today as well. If not what will tomorrows show consist of?
" @rentfn said:today's gonna be double and final jeopardy, i imagine tomorrow will be a regular match (reg, double and final)" I enjoyed it. I hope today's episode has more questions then back story about it. "It's seems that they are contnuing from yesterday and moving on to double jeopardy. Although, I wonder if they are going to do final jeopardy today as well. If not what will tomorrows show consist of? "
" Although, I wonder if they are going to do final jeopardy today as well. If not what will tomorrows show consist of? "They'll do Double Jeopardy! and Final Jeopardy! today, if I heard correctly. Tomorrow's going to be another complete game, presumably with less discussion of Watson's development, as this is supposed to be a two-game tournament-style exhibition match.
" I feel like I'm the only person in the world unimpressed with two people who have already been made fucking rich by Jeopardy coming back and being paid to take a dive in a series of 30 minute infomercials. *adjusts tin foil hat* "All winnings go to charity, unless you're referring to appearance fees.
" @fazzle said:No, he's clearly suggesting they're being paid to let Watson win." I feel like I'm the only person in the world unimpressed with two people who have already been made fucking rich by Jeopardy coming back and being paid to take a dive in a series of 30 minute infomercials. *adjusts tin foil hat* "All winnings go to charity, unless you're referring to appearance fees. "
" @fazzle said:Not entirely true, either of these. There's a $1,000,000 prize for winning, with $300,000 for second and $200,000 for third. While IBM will donate all their winnings to charity, both Ken and Brad are keeping half of their earnings, with the other half going to the charities of their choice." I feel like I'm the only person in the world unimpressed with two people who have already been made fucking rich by Jeopardy coming back and being paid to take a dive in a series of 30 minute infomercials. *adjusts tin foil hat* "All winnings go to charity, unless you're referring to appearance fees. "
Not only is there a good amount of money to win by beating Watson, but Brad Rutter's never lost on Jeopardy! before, and I'm sure Ken would love to win a match as prominent as this to extend his lifetime game show winnings (and exact a little revenge on Rutter for the Ultimate Tournament of Champions stomping he received). I highly doubt any of them would be willing to take a dive in the venue they excel at the most.
Yeah I've been watching the videos, and it's some crazy stuff. The fact that Watson can compete on Jeopardy in particular is impressive because the Jeopardy questions aren't always straight forward, often relying on word play or tricky puns for the layout. And things Watson gets incorrect are usually context based questions.
" @FrankCanada97 said:Yep this exactly. While I obviously have no way of proving this, and am simply going on baseless speculation, that's what I believe is happening. This entire thing is just to get Watson into the eye of the public, so that IBM can then go on to sell the technology to...let's say Microsoft, so they can have a draw to get the public away from Google and onto Bing." @fazzle said:No, he's clearly suggesting they're being paid to let Watson win. "" I feel like I'm the only person in the world unimpressed with two people who have already been made fucking rich by Jeopardy coming back and being paid to take a dive in a series of 30 minute infomercials. *adjusts tin foil hat* "All winnings go to charity, unless you're referring to appearance fees. "
I just heard about it today when they mentioned it on ESPN. I'll check it out tonight although I'm not sure how I feel about it. "Deep Blue" vs Kasparov in the 90's was one thing since it was all tactical probabilities. This time around it's a curiosity for certain, and those jokes about HAL or Skynet don't seem so remote.
Still, I think the whole thing would have been even more entertaining if the machine would have been competing against Sean Connery and "Turd Ferguson". Ha-ha!
Watson is absolutely creaming them right now. I am also baffled by the amount it chooses to wager. Also he "guessed" on a Double Jeopardy question.
Wow, Toronto? Really? Lester B. Pearson apparently led a double life and Billy Bishop was a WWII battle.
The ability of watson to read the question faster is a testament to its computing power.
Also changing the topic constantly seemed to have thrown the other two off
They didn't really get thrown off. The lower value questions are more directly worded so Watson was on a streak for a while. Once you get up to about 800 Watson can't quite understand most of the questions. So basically the contestants can let Watson hang itself by allowing it to buzz in to any question they think is too esoterically worded.
The answer for that Final Jeopardy question sort of proves Watson has some trouble with contextual questions? I really don't know how it could have gotten the country wrong on that one unless it was trying to tell a joke. Because if it was telling a joke than that means it's self aware, man! It knows! The second that thing gets connected to the internet we're all dead!
" The answer for that Final Jeopardy question sort of proves Watson has some trouble with contextual questions? I really don't know how it could have gotten the country wrong on that one unless it was trying to tell a joke. Because if it was telling a joke than that means it's self aware, man! It knows! The second that thing gets connected to the internet we're all dead! "Toronto is commonly compared to Chicago. Perhaps for some reason, it found that comparison in his data-banks and used it?
I'm intrigued, although I remember most of those categories and questions from regular Jeopardy. I watch Jeopardy like it's going out of style and I'm an old retired man.
I found an explanation for Watson's answer. Apparently, Chicago was a close second on the list." @FrankCanada97: True enough, but it still has trouble with the context of questions. It just seems to disregard certain bits of information (in this case the category "U.S. cities") and only provides the best categorical response. I'm impressed by it's analytical reasoning and deduction based upon the information it already knows, but feel a little safer inside knowing that some of it's logic isn't quite there yet. "
Link is here if you want to read more.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.
It's an impressive language parsing feat, but, like most Computer Science AI stuff, it probably wouldn't be so impressive if you knew how it worked. AI is like a magic trick in that way; its about making it seem like real intelligence.
People can't out run cars, wrestle a bulldozer or dig faster than a steam shovel -- but there's nothing meaningful or philosophical about that. That's how I feel about AI.
Please Log In to post.
Log in to comment