IBM's Watson on Jeopardy.

[Mise]

Half the people in this thread can't seem to understand language. One thing to NickyJ and his ilk's credit: they have indeed convinced me that the bar for considering a being "intelligent" is set far too low.

You're not being all cool and edgy and cynical, you're just being ignorant luddites, criticising technology that you don't even begin understand. That you simultaneously decry that same technology for "not understanding" things is downright hilarious.

 

[_random_]

Half the people in this thread can't seem to understand language. One thing to NickyJ and his ilk's credit: they have indeed convinced me that the bar for considering a being "intelligent" is set far too low.

You're not being all cool and edgy and cynical, you're just being ignorant luddites, criticising technology that you don't even begin understand. That you simultaneously decry that same technology for "not understanding" things is downright hilarious.

/thread

 

[Formaldehyde]

I think many people feel threatened by achievements such as this. They don't really understand how computers or other advanced technology works, so they develop defense mechanisms to try to cope with the unknown.

Any sufficiently advanced technology is indistinguishable from magic. Arthur C. Clarke

 

[sophie]

Still, if you ask Watson "Can you speak English?", it wouldn't be able to reply. It would only be able to say something along the lines of, "What is Henry the Eighth?"

This is also ignoring, of course, the fact that the computer also had to decide its confidence in its ability to answer the question, and decide to buzz in, as well as choosing daily double amounts based on confidence. It's actually a fairly remarkable machine, and aside from having ridiculous reaction times, and lack of nerves, really doesn't operate any differently than, say, Ken Jennings

 

[MeteorPunch]

Jeopardy isn't supposed to be quizbowl though, it's supposed to be Jeopardy. Simply because it requires anticipating when questions end, doesn't make it bad.

BTW the above is what makes Watson so impressive, he can accurately pick up on clues as to when the question is going to end.

Watson gets to know when the question ends, and gets first dibs at buzzing.

When host Alex Trebek finishes stating a clue, a human operator (who works for Jeopardy!) turns on a “Buzzer Enable” light on stage to indicate that contestants can “buzz in” and answer. At exactly the moment the “Buzzer Enable” light is activated, Watson’s system receives a signal that the buzzer is open.

 

[MeteorPunch]

Well that's easily and demonstratably wrong, not just by looking at the programming but by testing if his answers line up with a google search.
For example when given:
"Before this hotel mogul's elbow broke through it, a Picasso he owned was worth $139 million; after, $85 million"
It answered "Steven Wynn." Google answered "Wynn Vs. Lloyd's of London Over Damaged Picasso."
When given:
"This town is known as "Sin City" & its downtown is "Glitter Gulch""
It answered "Las Vegas" Google answered "The Nugget and the Rest of Benny's Corner - Downtown Glitter Gulch"

You think they are using Google's exact search method?

Then Google needs to GIT OUT of the USA.

Sigworthy.

Half the people in this thread can't seem to understand language. One thing to NickyJ and his ilk's credit: they have indeed convinced me that the bar for considering a being "intelligent" is set far too low.

You're not being all cool and edgy and cynical, you're just being ignorant luddites, criticising technology that you don't even begin understand. That you simultaneously decry that same technology for "not understanding" things is downright hilarious.

I'm more disappointed that it isn't magic. It's a nice accomplishment, but throwing it on Jeopardy while using completely different of rules that the human players is disingenuous. It's a powerful search engine, but grandmas at home are thinking it's the first T-1000.

 

[illram]

Was real money on the line in this game, or did this game "count?" (Isn't there some Jeopardy "rank" system where people's wins and losses are tabulated?) I am assuming not... it would be pretty messed up if your unbeaten streak is put on the line in the first ever battle vs. your new Supercomputer Robot Overlord.

Either way this is awesome and I only hope things like this spark more research and money into better AI.

 

[SuperJay]

As I understand it, Jennings and Sutter (sp?) kept half their winnings, with the other half donated to the charity of their choice. IBM donated all of Watson's winnings to charity.

And yeah, what Mise said. I'm not even sure what there is to be "against" on this topic, but that never stops folks from arguing.... something.

 

[NickyJ]

As I understand it, Jennings and Sutter (sp?) kept half their winnings, with the other half donated to the charity of their choice. IBM donated all of Watson's winnings to charity.

Rutter, if I remember correctly. Anyway, I thought it was said that all of them would donate all of their winnings to charity, but I don't know for sure.

 

[warpus]

The IBM team responsible for the creation of Watson recently answered a bunch of community created questions on reddit. Here's some excerpts:

2. What was the biggest technological hurdle you had to overcome in the development of Watson? (this_is_not_the_cia)

Accelerating the innovation process – making it easy to combine, weigh evaluate and evolve many different independently developed algorithms that analyze language form different perspectives.

Watson is a leap in computers being able to understand natural language, which will help humans be able to find the answers they need from the vast amounts of information they deal with everyday. Think of Watson as a technology that will enable people to have the exact information they need at their fingertips

3. Can you walk us through the logic Watson would go through to answer a question such as, "The antagonist of Stevenson's Treasure Island." (Who is Long John Silver?) (elmuchoprez)

Step One: Parses sentence to get some logical structure describing the answer
X is the answer.
antagonist(X).
antagonist_of(X, Stevenson's Treasure Island).
modifies_possesive(Stevenson, Treasure Island).
modifies(Treasure, Island)

Step Two: Generates Semantic Assumptions
island(Treasure Island)
location(Treasure Island)
resort(Treasure Island)
book(Treasure Island)
movie(Treasure Island)

person(Stevenson)
organization(Stevenson)
company(Stevenson)
author(Stevenson)
director(Stevenson)

person(antagonist)
person(X)

Step Three: Builds different semantic queries based on phrases, keywords and semantic assumptions.

Step Four: Generates 100s of answers based on passage, documents and facts returned
from 3. Hopefully Long-John Silver is one of them.

Step Five: For each answer formulates new searches to find evidence in support or
refutation of answer -- score the evidence.
Positive Examples:
Long-John Silver the main character in Treasure Island.....
The antagonist in Treasure Island is Long-John Silver
Treasure Island, by Stevenson was a great book.
One of the great antagonists of all time was Long-John Silver
Richard Lewis Stevenson's book, Treasure Island features many great
characters, the greatest of which was Long-John Silver.

Step Six: Generate, get evidence and score new assumptions
Positive Examples: (negative examples would support other characters,
people, books, etc associated with any Stevenson, Treasure or Island)
Stevenson = Richard Lewis Stevenson
"by Stevenson" --> Stevenson's
main character --> antagonist

Step Seven: Combine all the evidence and their scores
Based on analysis of evidence for all possible answer compute a final
confidence and link back to the evidence.
Watson's correctness will depend on evidence collection, analysis and
scoring algorithms and the machine learning used to weight and combine the
scores.

Maybe this will shut up the conspiracy theorists.. then again probably not.

 

[Formaldehyde]

Wait a minute here! Isn't this what reasonably intelligent humans naturally do without even thinking about it when they use google to find answers?

 

[bhsup]

I won't be impressed until Watson can beat The Almighty.

 

[Alassius]

Half the people in this thread can't seem to understand language. One thing to NickyJ and his ilk's credit: they have indeed convinced me that the bar for considering a being "intelligent" is set far too low.

You might be interested in these topics:

http://en.wikipedia.org/wiki/Strong_AI
http://en.wikipedia.org/wiki/Turing_test
http://en.wikipedia.org/wiki/Chinese_room

Basically, whether a machine that appears to be intelligent should be considered intelligent is a very old question, about as old as computer itself. The "Strong AI" school says not until it can actually think like a human, whereas the "Weak AI" school says it's intelligent as long as you can't tell if you are talking to a human or a computer (i.e. Turing test).

The general consensus in the sixties was that the Strong AI was just around the corner. The general consensus today is that Weak AI (in very restricted fields) is just around the corner, and we don't know if Strong AI is possible.

 

[duckstab]

Is it too early to get the Butlerian Jihad started?

 

[NickyJ]

The IBM team responsible for the creation of Watson recently answered a bunch of community created questions on reddit. Here's some excerpts:

Watson is a leap in computers being able to understand natural language, which will help humans be able to find the answers they need from the vast amounts of information they deal with everyday. Think of Watson as a technology that will enable people to have the exact information they need at their fingertips

First off, good on them for blowing their own horns. Can't get enough of that. Anyway:

Oh, goodie! We get another Google search, but from IBM! I was never able to "find the answers" I "need from the vast amounts of information" I "deal with everyday" with Google.

 

[Haseri]

The IBM team responsible for the creation of Watson recently answered a bunch of community created questions on reddit. Here's some excerpts:



Maybe this will shut up the conspiracy theorists.. then again probably not.

Pfft, like conspiracy theorists will let trivial things like facts get in the way of the Truth.

@NickyJ Yes, but Google needs hundreds of server farms. Watson only needs a room full of computers.

So yes, it is rather like Google, except it isn't hooked up to the internet, is smaller and generates concise answers instead of web pages where I might find the answer.

And come on, you've never had pride in your work before?

 

[warpus]

Wait a minute here! Isn't this what reasonably intelligent humans naturally do without even thinking about it when they use google to find answers?

Yep, because our brains are incredibly amazing pattern recognition machines that allow us to process natural language and find meaning in it relatively easily.. something that is incredibly hard to get a computer to do.

So yes, it is rather like Google, except it isn't hooked up to the internet, is smaller and generates concise answers instead of web pages where I might find the answer.

And come on, you've never had pride in your work before?

Yes, but also.. google doesn't do any natural language processing. Watson does. That's a big part of what it does, actually, the searching/googling is elementary, really.

 

[ParkCungHee]

So yes, it is rather like Google, except it isn't hooked up to the internet, is smaller and generates concise answers instead of web pages where I might find the answer.

It is rather like Google in that an Automobile is a bit like a Jet Plane.
Google simply matches like to like, then prioritizes based on the number of times something has been linked to.
Specific wording can throw things off entirely, and there are no linkages made between the words at all.
Watson is able to actually break down human sentences to understand meaning behind them. They are similar only in that they are programs that are (currently) used to conduct information searches.
Even that is unlikely to last long.
The principals behind Watson are likely to be applied to computers in a variety of other roles soon. For example, if I was to say "Shut down all the trash compactors on the detention level!" Google can only know you want something that involves the words Compactors, Detention, Level, Shut, Trash, Down, All etc.
Watson can actually break that sentence down into it's meaning as a statement and act based on that statement.

It really is a marvelous step forward.

 

[Formaldehyde]

Yep, because our brains are incredibly amazing pattern recognition machines that allow us to process natural language and find meaning in it relatively easily.. something that is incredibly hard to get a computer to do.

Indeed. You apparently missed my sarcasm and even have me confused with the Luddites who obviously don't know much about computers, which should be rather obvious given all my previous responses in this thread.

Yes, but also.. google doesn't do any natural language processing. Watson does. That's a big part of what it does, actually, the searching/googling is elementary, really.

Um, that wasn't me who stated that. I would appreciate it if you edited your post to reflect the person who actually stated it.

 

[Dachs]

Wait a minute here! Isn't this what reasonably intelligent humans naturally do without even thinking about it when they use google to find answers?

Yes, the problem with computational linguistics has never really been designing the searches to imitate human algorithms (remains to be seen whether those are the most efficient or not), it's been optimization, and I'm semi-sure that meaningful (not hardware-based) optimization is NP-hard. It's been a few years, though, so my recollections might be a bit rusty?