Is Watson the most advanced super computer in existence?

[Lindley]

99% of the human population can't sort 10,000 files into alphabetical order in less than a second either, at least by hand.

Computers can. That doesn't mean they're intelligent, it just means someone figured out a fast algorithm to accomplish that task.

On the other hand, it will take a very, very long time until, for example, a computer beats a human in visual recognition. Because that's what the human brain can do with perfection.

Even that will come in a matter of years, decades at most. Visual recognition is a solvable problem with enough math and computing power.

Machine self-awareness is an entirely different category. It's not clear whether that will ever be possible at this point.

 

[CorporalCaptain]

But Jeopardy is not about intelligence, it's about having vast amounts of trivia knowledge. Pretty much anyone can accumulate that with some time and effort. It's not a question of intellectual ability.

I don't want to sound like I'm undermining your point here, because I generally agree with it. However, there is an application of problem solving in interpreting the questions (I mean answers) posed by Alex, in order to come up with the correct response. The "answers" Alex gives often involve some kind of pun, or other word play or loose association. Interpreting the word play correctly can often point the way to making an educated guess, when you don't know the correct response (something that Watson seemed to fail miserably at). In that way, at least, playing Jeopardy well does involve some application of intelligence.

 

[JarodRussell]

99% of the human population can't sort 10,000 files into alphabetical order in less than a second either, at least by hand.

Computers can. That doesn't mean they're intelligent, it just means someone figured out a fast algorithm to accomplish that task.

On the other hand, it will take a very, very long time until, for example, a computer beats a human in visual recognition. Because that's what the human brain can do with perfection.

Even that will come in a matter of years, decades at most. Visual recognition is a solvable problem with enough math and computing power.

Machine self-awareness is an entirely different category. It's not clear whether that will ever be possible at this point.

It's not impossible, sure. But a computer the size of a human head being able to do visual recognition tasks as good as the average human, that's a long time away. While the tiniest pocket calculator can perform calculations can outperform any human.

As said above, Watson might be a cool thing, but it also takes 750 servers. A huge space is currently needed for that. Of course this will change, no doubt about that.

 

[DarthTom]

On the other hand, it will take a very, very long time until, for example, a computer beats a human in visual recognition. Because that's what the human brain can do with perfection.

I thought the NSA has made great strides in facial recognition software that can scan crowds for terror suspects which also concerns privacy rights advocates?

Facial Recognition software NSA is working on

 

[JarodRussell]

On the other hand, it will take a very, very long time until, for example, a computer beats a human in visual recognition. Because that's what the human brain can do with perfection.

I thought the NSA has made great strides in facial recognition software that can scan crowds for terror suspects which also concerns privacy rights advocates?

Facial Recognition software NSA is working on

Yeah, but a human can still absolutely outperform this. If you know the target, you can pick it out of a crowd with ease. And can do so much more than just recognize faces at the same time.

The current advantage here is automization. You don't need humans who get tired watching computer monitors.

 

[Lindley]

On the other hand, it will take a very, very long time until, for example, a computer beats a human in visual recognition. Because that's what the human brain can do with perfection.

I thought the NSA has made great strides in facial recognition software that can scan crowds for terror suspects which also concerns privacy rights advocates?

Facial Recognition software NSA is working on

Face detection is coming along, and face recognition is just a standard machine learning problem once you've detected the face.

The ability to recognize specific *types* of things is eminently doable with enough training data. The ability to generalize that to recognizing anything you've seen before, or figure out the purpose of something you haven't seen before, is still a bit beyond current technology.

 

[DarthTom]

I did some addtional reserach and Watson isn't the most advanced computer but he's up there. I had never heard of this system but Univeristy of Illinois tested 'Concept Net 4' developed by MIT. They found it has the intelliegence of a 4 year old.

WASHINGTON: One of the world's best artificial intelligence computers is only as smart as a four year old human child, a new study suggests. Researchers at the University of Illinois at Chicago IQ-tested one of the best available artificial intelligence systems to see how intelligent it really is.
It turned out that the computer is about as smart as the average four-year-old, researchers said.
The UIC team put ConceptNet 4, an artificial intelligence system developed at Massachusetts Institute of Technology (MIT), through the verbal portions of the Weschsler Preschool and Primary Scale of Intelligence Test, a standard IQ assessment for young children.
They found ConceptNet 4 has the average IQ of a young child. But unlike most children, the machine's scores were very uneven across different portions of the test.
"If a child had scores that varied this much, it might be a symptom that something was wrong," said Robert Sloan, professor and head of computer science at UIC, and lead author on the study.
Sloan said ConceptNet 4 did very well on a test of vocabulary and on a test of its ability to recognise similarities.
"But ConceptNet 4 did dramatically worse than average on comprehension - the 'why' questions," he said.
One of the hardest problems in building an artificial intelligence, Sloan said, is devising a computer programme that can make sound and prudent judgement based on a simple perception of the situation or facts-the dictionary definition of commonsense.
Commonsense has eluded artificial intelligence engineers because it requires both a very large collection of facts and what Sloan calls implicit facts-things so obvious that we don't know we know them. A computer may know the temperature at which water freezes, but we know that ice is cold.
"All of us know a huge number of things. As babies, we crawled around and yanked on things and learned that things fall. We yanked on other things and learned that dogs and cats don't appreciate having their tails pulled," said Sloan.
"We are still very far from programmes with commonsense - artificial intelligence that can answer comprehension questions with the skill of a child of eight," said Sloan.

Google is also apparently using a Quantum Computer for AI like modeling:

T
he nonprofit Universities Space Research Association just splurged on a new toy: A $10 million D-Wave Two quantum computer. The machine is the most advanced commercially available quantum computer in the world.
USRA installed the device, which should be online by the end of September, at the new Quantum Artificial Intelligence Lab at the Ames Research Center in Silicon Valley, where researchers from USRA, NASA, and Google will put it use.
What is a quantum computer? Think of it this way: Regular computers process information using bits, which can innately have only one of two values (zero or one, plus or minus, yes or no, etc). Quantum computers use qubits, which can simultaneously represent more than one value. In that sense, "quantum computing could herald a new era of number-crunching," said Robert McMillan at Wired, by breaking out of the "binary" system of computing. USA Today explains:
A quantum computer takes advantage of a principle in physics that on the atomic scale particles can be in many places at once or have multiple properties such as magnetism or an electrical charge simultaneously. They would store data in these multiple positions at once, unlike conventional computers, which only store one value or data point in their circuits at a time. [USA Today]
That means Qubits can multitask, which results in outstanding power. "A quantum computer with 300 qubits could run more calculations in an instant than there are atoms in the universe," says Technology Review.

 

[JarodRussell]

There also seems to be a bit of confusion between computers and software. Watson is not a super computer. It's a piece of software, running on a server farm. And that farm isn't even that big.

Here's a list of actual supercomputers as of November 2013.

http://www.top500.org/lists/2013/11/

 

[iguana_tonante]

Is this forum not where science and technology is discussed? Do you have anything to contribute to the conversation?

Yes. Do you?

So yes - sorry - super computing is exciting to me and I don't understand what happens under the proverbial hood of the car so I ask questions.

It's not that you are not particularly knowledgeable about computing. It's a pretty obscure field. It's that you don't seem to be reading what people are saying here.

Any computer that can beat a human being at Jeporday - IMO - is an amazing feat considering that 99% of the human population could not beat the flesh and blood winners either.

QED.

Here's a list of actual supercomputers as of November 2013.

http://www.top500.org/lists/2013/11/

I have access to #15.

 

[Gov Kodos]

I have access to #15.

Thanks, Bender.

 

[scotpens]

No. Watson is not the "most advanced" super computer on the planet. It's just a really good search engine, with good publicity.

Meh. I don't think you're giving, "him," enough credit. How many computers do you run accross in everyday life that can beat not 1 but 2 or 3 Jeporady champions using natural language interaction?

There are computers that can beat the best human players at chess. That doesn't mean the computer is intelligent. It's just capable of calculating the outcomes of thousands of possible sequences of moves really, really fast.

 

[RAMA]

Watson is open source and on the cloud now:

http://www.kurzweilai.net/ibm-to-take-watson-to-the-cloud-opens-to-app-developers

November 14, 2013



A hypothetical Watson medical health app (credit: IBM)

IBM announced today that it will make its IBM Watson technology available to developers in the cloud so they can build apps using Watson.
IBM will be launching the IBM Watson Developers Cloud, a cloud-hosted marketplace for resources including a developer toolkit, educational materials, and access to Watson’s application programming interface (API).

 

[Asbo Zaprudder]

No. Watson is not the "most advanced" super computer on the planet. It's just a really good search engine, with good publicity.

Meh. I don't think you're giving, "him," enough credit. How many computers do you run accross in everyday life that can beat not 1 but 2 or 3 Jeporady champions using natural language interaction?

There are computers that can beat the best human players at chess. That doesn't mean the computer is intelligent. It's just capable of calculating the outcomes of thousands of possible sequences of moves really, really fast.

"Asking whether a computer can think is like asking if a submarine can swim." — Edsger W. Dijkstra

We're always going to find ways to think ourselves superior. Let's hope the machines don't do likewise.

 

[DarthTom]

A hypothetical Watson medical health app (credit: IBM)

IBM announced today that it will make its IBM Watson technology available to developers in the cloud so they can build apps using Watson.
IBM will be launching the IBM Watson Developers Cloud, a cloud-hosted marketplace for resources including a developer toolkit, educational materials, and access to Watson’s application programming interface (API).

That's very cool. I wonder if it will ever become as ubiquitous as Siri?

I think they should place Watson software in public libraries. It may generate interest in school children visiting libraries again.

 

[Amaris]

But Jeopardy is not about intelligence, it's about having vast amounts of trivia knowledge. Pretty much anyone can accumulate that with some time and effort. It's not a question of intellectual ability.

I don't want to sound like I'm undermining your point here, because I generally agree with it. However, there is an application of problem solving in interpreting the questions (I mean answers) posed by Alex, in order to come up with the correct response. The "answers" Alex gives often involve some kind of pun, or other word play or loose association. Interpreting the word play correctly can often point the way to making an educated guess, when you don't know the correct response (something that Watson seemed to fail miserably at). In that way, at least, playing Jeopardy well does involve some application of intelligence.

tch. Any guy can Jeopardy good. Can you add a jillion plus a jillion? Can't explain that!

 

[rhubarbodendron]

my appologies if I make a point that perhaps was already mentioned. I must admit I didn't read the whole thread since I have a pretty bad cold and it's getting close to bedtime over here.

What I would like to add to the discussion is:

Can a computer ever excel human intelligence? (Cynicists may ask: is there any such thing at all but let's leave that aside atm, please)

There's an old IT proverb "a computer is always as smart as two people: its programmer and its user."
That would imply that a computer can not be smarter than his programmer.
Or could it? In that case the programmer would have to program the computer to be smarter than himself. But how does one do that??

(Just so that you know what you might perhaps have to explain to me in greater detail here are my "personal specifications"
My own experience with programming is both extensive (regarding the timespan) and limited (as far as my abilities are concerned). I started way back in the 70s with BASIC and over the course of 40 years worked my way up to PHP and HTML. Programming in the modern languages doesn't hold much appeal to me, though. I much prefer the more direct contact to the computer that BASIC offered. I have no immediate experience with fuzzy logic and systems that have an ability to learn unless you count that I was one of Alice's many teachers and training partners.)

 

[JarodRussell]

Or could it? In that case the programmer would have to program the computer to be smarter than himself. But how does one do that??

More programmers. And/or the ability to learn.

 

[Robert Maxwell]

It all comes down to the definition of "intelligence." Computers can do things easily that humans may find difficult or even impossible. And yet, were you to show a computer a painting, it wouldn't be able to analyze it the way an art critic could. Computers can crunch numbers exceptionally well, but they have no critical capacity.

 

[JarodRussell]

It all comes down to the definition of "intelligence." Computers can do things easily that humans may find difficult or even impossible. And yet, were you to show a computer a painting, it wouldn't be able to analyze it the way an art critic could. Computers can crunch numbers exceptionally well, but they have no critical capacity.

But all of this is a big "NOT YET". All it takes is the right processing power and programming.

And what we want from a computer. Computers are (usually) supposed to deliver deterministic results. You want it to function according to your needs. And art critic is quite the opposite of that. Show one image to two critics and you get three different opinions. On a Monday. On a Tuesday, you get an additional set of opinions.

Right now, you can write programs to objectively analyze an image, and depending on the processing power you get amazing results really fast. The subjective interpretation is usually not what you want a computer to do. Even today you could write a program that spits out "subjective" analysis, depending on a database of opinions/rules/heuristics. Like, I dunno, blue color - cold, orange color - warm, and you can yield a result like "this image has a cold feel to it". As a basic example. But what is the point of such a program?

 

[Robert Maxwell]

It all comes down to the definition of "intelligence." Computers can do things easily that humans may find difficult or even impossible. And yet, were you to show a computer a painting, it wouldn't be able to analyze it the way an art critic could. Computers can crunch numbers exceptionally well, but they have no critical capacity.

But all of this is a big "NOT YET". All it takes is the right processing power and programming.

And what we want from a computer. Computers are (usually) supposed to deliver deterministic results. You want it to function according to your needs. And art critic is quite the opposite of that. Show one image to two critics and you get three different opinions. On a Monday. On a Tuesday, you get an additional set of opinions.

Right now, you can write programs to objectively analyze an image. The subjective interpretation is usually not what you want a computer to do.

Computers don't always have to behave deterministically. (Some are designed not to, for certain applications.)

Computers can do quantitative analysis quite well. Even when they play at qualitative analysis, in the end it's just a bunch of rules surrounding a quantitative analysis.

For most purposes, though, you're right: most problems we can envision will eventually be solvable by a computer that is powerful enough and programmed properly. The latter depends on having good enough programmers, of course. The most powerful computer in the world won't help you solve any problems if it doesn't have the right programmers.