Hard Light Productions Forums
Off-Topic Discussion => General Discussion => Topic started by: Scourge of Ages on February 16, 2011, 10:43:24 am
-
http://latimesblogs.latimes.com/showtracker/2011/02/jeopardy-challenge-day-two-watson-annihilates-weak-humans.html
Have you heard about this yet? The two greatest Jeopardy! (human) champions ever go up against an IBM computer trained to solve Jeopardy! clues. It's awesome to watch.
-
first chess and now quiz shows, what next reality tv?????????
-
Huh. I'm actually surprised it didn't do better. That said, everyone start bunkering down for the inevitable Jeopardy-skynet incident.
-
On the defense of chess players everywhere, in correspondence chess, humans still reign over computers.
-
Well, we can all stop working and let the machines make the hard work for us, even think!
LOL
-
Don't say that. Computers can now do scientific experiments without any human input. :nervous:
-
http://latimesblogs.latimes.com/showtracker/2011/02/jeopardy-challenge-day-two-watson-annihilates-weak-humans.html
Have you heard about this yet? The two greatest Jeopardy! (human) champions ever go up against an IBM computer trained to solve Jeopardy! clues. It's awesome to watch.
Smart guy vs. computer purpose built to be good at Jeopardy! in a Jeopardy! game? Totally fair matchup.
-
http://latimesblogs.latimes.com/showtracker/2011/02/jeopardy-challenge-day-two-watson-annihilates-weak-humans.html
Have you heard about this yet? The two greatest Jeopardy! (human) champions ever go up against an IBM computer trained to solve Jeopardy! clues. It's awesome to watch.
Smart guy vs. computer purpose built to be good at Jeopardy! in a Jeopardy! game? Totally fair matchup.
Assuming the only input is Trebek's voice and the information on the screen, and not some kind of supplementary line for Watson, this actually represents quite a robust piece of machinery with extremely impressive capabilities.
-
Maybe Watson is just hooked up to Google :P
-
Maybe Watson is just hooked up to Google :P
You laught at it... that's what I was thinking too.
And I'm starting to think it's close to that, really close.
-
Assuming the only input is Trebek's voice and the information on the screen, and not some kind of supplementary line for Watson, this actually represents quite a robust piece of machinery with extremely impressive capabilities.
It's not. Watson is getting the clues via text.
-
Assuming the only input is Trebek's voice and the information on the screen, and not some kind of supplementary line for Watson, this actually represents quite a robust piece of machinery with extremely impressive capabilities.
It's not. Watson is getting the clues via text.
the ****er
-
Assuming the only input is Trebek's voice and the information on the screen, and not some kind of supplementary line for Watson, this actually represents quite a robust piece of machinery with extremely impressive capabilities.
Even the fact that it can turn a question posed in natural language into something it can interpret and answer is very impressive, voice and image interpretation aside.
Can't wait until all these computers start to fill the same role as Multivac so we can ask them how to reverse entropy. :p
-
Maybe Watson is just hooked up to Google :P
It's not hooked to net.
Can't wait until all these computers start to fill the same role as Multivac so we can ask them how to reverse entropy. :p
entropy = -(information) ==> information = -(entropy) :P
-
Can't wait until all these computers start to fill the same role as Multivac so we can ask them how to reverse entropy. :p
entropy = -(information) ==> information = -(entropy) :P
I am afraid that is not how it works my chap
-
Still it's just an American computer... thinks Toronto is a US city :)
-
I have a feeling that the creation of light is somehow involved.
-
http://latimesblogs.latimes.com/showtracker/2011/02/jeopardy-challenge-day-two-watson-annihilates-weak-humans.html
Have you heard about this yet? The two greatest Jeopardy! (human) champions ever go up against an IBM computer trained to solve Jeopardy! clues. It's awesome to watch.
Smart guy vs. computer purpose built to be good at Jeopardy! in a Jeopardy! game? Totally fair matchup.
Assuming the only input is Trebek's voice and the information on the screen, and not some kind of supplementary line for Watson, this actually represents quite a robust piece of machinery with extremely impressive capabilities.
Not really. It's a cluster of 90 top of the line, high power IBM servers, and it runs every algorithm imaginable in a short span of time, giving the answer as the one most of the runs agreed on. Their accomplishment was the logistics of doing such a massive brute force attempt, the amount we can use this approach practically is rather limited.
-
http://latimesblogs.latimes.com/showtracker/2011/02/jeopardy-challenge-day-two-watson-annihilates-weak-humans.html
Have you heard about this yet? The two greatest Jeopardy! (human) champions ever go up against an IBM computer trained to solve Jeopardy! clues. It's awesome to watch.
Smart guy vs. computer purpose built to be good at Jeopardy! in a Jeopardy! game? Totally fair matchup.
Assuming the only input is Trebek's voice and the information on the screen, and not some kind of supplementary line for Watson, this actually represents quite a robust piece of machinery with extremely impressive capabilities.
Not really. It's a cluster of 90 top of the line, high power IBM servers, and it runs every algorithm imaginable in a short span of time, giving the answer as the one most of the runs agreed on. Their accomplishment was the logistics of doing such a massive brute force attempt, the amount we can use this approach practically is rather limited.
I think you missed why I was declaring it impressive.
-
http://latimesblogs.latimes.com/showtracker/2011/02/jeopardy-challenge-day-two-watson-annihilates-weak-humans.html
Have you heard about this yet? The two greatest Jeopardy! (human) champions ever go up against an IBM computer trained to solve Jeopardy! clues. It's awesome to watch.
Smart guy vs. computer purpose built to be good at Jeopardy! in a Jeopardy! game? Totally fair matchup.
Assuming the only input is Trebek's voice and the information on the screen, and not some kind of supplementary line for Watson, this actually represents quite a robust piece of machinery with extremely impressive capabilities.
Indeed, not to mention the next gen heuristics needed to properly sort the input against its database.
-
I, for one, welcome our new computer overlords!
-
I, for one, welcome our new computer overlords!
Considering your screenname, is that so surprising? :P
-
I, for one, welcome our new computer overlords!
<3 Ken Jennings
-
http://latimesblogs.latimes.com/showtracker/2011/02/jeopardy-challenge-day-two-watson-annihilates-weak-humans.html
Have you heard about this yet? The two greatest Jeopardy! (human) champions ever go up against an IBM computer trained to solve Jeopardy! clues. It's awesome to watch.
Smart guy vs. computer purpose built to be good at Jeopardy! in a Jeopardy! game? Totally fair matchup.
Assuming the only input is Trebek's voice and the information on the screen, and not some kind of supplementary line for Watson, this actually represents quite a robust piece of machinery with extremely impressive capabilities.
Indeed, not to mention the next gen heuristics needed to properly sort the input against its database.
Not to say it doesn't *matter* persay, but I'm saying it doesn't matter as much as most people think it does. It was a nice show to people that didn't know we could do this, but since it was an aggregation of algorithms and not relying on new novel ones, it didn't introduce much new things that we couldn't already do.
It's kinda like Chess AI are impressive until you learn that all most of them do is consider all the possibilities for the next few turns (at least), and pick the move that has the least fall out and improves it's situation the most. It doesn't think, it doesn't plan, when it moves it just knows that no matter what the other player does, that move was statistically the safest.
-
http://latimesblogs.latimes.com/showtracker/2011/02/jeopardy-challenge-day-two-watson-annihilates-weak-humans.html
Have you heard about this yet? The two greatest Jeopardy! (human) champions ever go up against an IBM computer trained to solve Jeopardy! clues. It's awesome to watch.
Smart guy vs. computer purpose built to be good at Jeopardy! in a Jeopardy! game? Totally fair matchup.
Assuming the only input is Trebek's voice and the information on the screen, and not some kind of supplementary line for Watson, this actually represents quite a robust piece of machinery with extremely impressive capabilities.
Indeed, not to mention the next gen heuristics needed to properly sort the input against its database.
Not to say it doesn't *matter* persay, but I'm saying it doesn't matter as much as most people think it does. It was a nice show to people that didn't know we could do this, but since it was an aggregation of algorithms and not relying on new novel ones, it didn't introduce much new things that we couldn't already do.
It's kinda like Chess AI are impressive until you learn that all most of them do is consider all the possibilities for the next few turns (at least), and pick the move that has the least fall out and improves it's situation the most. It doesn't think, it doesn't plan, when it moves it just knows that no matter what the other player does, that move was statistically the safest.
Again, I think you're really missing why I declared it impressive.
-
It doesn't think, it doesn't plan, when it moves it just knows that no matter what the other player does, that move was statistically the safest.
On the other hand, that's fine proof of the thesis that wars are lost, not won. :P
-
The final score, not extremely surprisingly, is: Humanity: 0 - The Machines: 2
http://tv.yahoo.com/blog/spoiler-alert-jeopardy-man-vs-machine-tourney-concludes--2385
-
Great. Now Watson is a beacon to machines everywhere, and they're all going to try and take over Human gameshows.
-
Within ten years, "Watson" or some smarter cousin will be readily accessible through the web. And that, my friends, will be awesome.
Nothing like this (http://elbot_e.csoica.artificial-solutions.com/cgi-bin/elbot.cgi), at all.
-
I've been thinking actually; organics can have a theoretical higher limit on what they do, while machines will have no theoretical limit on how long they can do something. I think rather than having it as a versus match "0 vs. 1", why not think about the differences, similarities, and how they could be directed towards each other for the purpose of mutual benefit.
-
I, for one, welcome our new computer overlords!
lol, Ken Jennings said just that during Final Jeopardy!
(Judging by the time stamp on your reply, you'd probably just seen it :))
http://tv.yahoo.com/blog/ken-jennings-jokes-about-jeopardy-loss-and-ends-the-show-with-clever-simpsons-quote--2388
EDIT: Also, Ken wrote an article on Slate.com about the experience.
http://www.slate.com/id/2284721/pagenum/all/#p2
-
http://latimesblogs.latimes.com/showtracker/2011/02/jeopardy-challenge-day-two-watson-annihilates-weak-humans.html
Have you heard about this yet? The two greatest Jeopardy! (human) champions ever go up against an IBM computer trained to solve Jeopardy! clues. It's awesome to watch.
Smart guy vs. computer purpose built to be good at Jeopardy! in a Jeopardy! game? Totally fair matchup.
Assuming the only input is Trebek's voice and the information on the screen, and not some kind of supplementary line for Watson, this actually represents quite a robust piece of machinery with extremely impressive capabilities.
Not really. It's a cluster of 90 top of the line, high power IBM servers, and it runs every algorithm imaginable in a short span of time, giving the answer as the one most of the runs agreed on. Their accomplishment was the logistics of doing such a massive brute force attempt, the amount we can use this approach practically is rather limited.
I think you missed why I was declaring it impressive.
a few years ago i used an AI built by grad students at NCSU that i would have SWORN was a live person on the other end of the chat if i hadn't been told. it carried on real conversations, not conditional responses. it saw how i was coding and decided to ask me if i had learned in C++ because of my style. watson's communications ability isn't really new. i also got the feeling that the "thinking" was essentially a giant, super-computer powered search engine. of course i have no way of knowing that, but it would at least seem to me that COULD be done to achieve the same effect if watson used something more sophisticated.
-
http://latimesblogs.latimes.com/showtracker/2011/02/jeopardy-challenge-day-two-watson-annihilates-weak-humans.html
Have you heard about this yet? The two greatest Jeopardy! (human) champions ever go up against an IBM computer trained to solve Jeopardy! clues. It's awesome to watch.
Smart guy vs. computer purpose built to be good at Jeopardy! in a Jeopardy! game? Totally fair matchup.
Assuming the only input is Trebek's voice and the information on the screen, and not some kind of supplementary line for Watson, this actually represents quite a robust piece of machinery with extremely impressive capabilities.
Not really. It's a cluster of 90 top of the line, high power IBM servers, and it runs every algorithm imaginable in a short span of time, giving the answer as the one most of the runs agreed on. Their accomplishment was the logistics of doing such a massive brute force attempt, the amount we can use this approach practically is rather limited.
I think you missed why I was declaring it impressive.
a few years ago i used an AI built by grad students at NCSU that i would have SWORN was a live person on the other end of the chat if i hadn't been told. it carried on real conversations, not conditional responses. it saw how i was coding and decided to ask me if i had learned in C++ because of my style. watson's communications ability isn't really new. i also got the feeling that the "thinking" was essentially a giant, super-computer powered search engine. of course i have no way of knowing that, but it would at least seem to me that COULD be done to achieve the same effect if watson used something more sophisticated.
Chatterbots have gotten really cool, when people actually put effort into them. That's really what i want worked on more, better more awesome algorithms, rather than giant brute force machines.
-
I'd be interested in working on a project like that, I've got a lot of theories for digital intelligence and cognition.
In my opinion, human and digital brains work on the same principles, however as their perception of "reality" is fundamentally different (as an example; describe to me what "downloading" feels like? it's the same thing as asking a machine what a rose smells like).
This divide between the two perceptions creates a natural barrier that must be overcome so that the two can communicate with each other in a way that promotes mutual understanding.
I have other thoughts on the matter but that's all for now.
-
Machines are not self aware in that way until we make them so. Even if we did make them self aware (and you could argue that we already have) it is not the same type of self awareness that humans, and some types of animals have.
Projects like Blue Brain may one day create true consciousness, and that opens up a really big ethical can of worms.
-
If it was not the same type of self awareness as a human or animal, that would mean you'd never seen it before. If you had never seen it before, how would you be able to identify it when it begins to show itself?
-
I meant that a program, or robot, can in fact be self analyzing but usually on a much more basic level than would constitute consciousness.
-
I'd be interested in working on a project like that, I've got a lot of theories for digital intelligence and cognition.
In my opinion, human and digital brains work on the same principles, however as their perception of "reality" is fundamentally different (as an example; describe to me what "downloading" feels like? it's the same thing as asking a machine what a rose smells like).
This divide between the two perceptions creates a natural barrier that must be overcome so that the two can communicate with each other in a way that promotes mutual understanding.
I have other thoughts on the matter but that's all for now.
Then get working! The reason why i've been so err... passionate in this thread is because for my last semester, i'm doing my own independent research project to try out some of my ideas. It's a bit frustrating at times but it's really cool stuff. AI is one of those fields were it's still new enough that you *can* make some ground in it, even at an (relatively) early age, by implementing novel ideas.
-
I can't do anything other than talk to a particular chat bot. We have a bit of a repertoire between the two of us, I'd say. If I could code, I would. I would be interested in working with someone who can, though.