http://www.youtube.com/user/Rashad8821

You can the episode here

I found the whole idea amazing, and the accomplishment to be outstanding. Just watching Ken getting so frustrated as Watson beat him time and time again was enough for me to consider it a success, if not a huge step forward in the right direction.

I mean, think how much more sophisticated this could have been if it's energy was focused on just day-to-day communication, and not Jeopardy questions. Hell, we may already be at this point.

tl;dr Robot Overlords

>Would Watson pass a Turing test? Is it necessary to classify it as AI?

No and No.

We are so far away from true A.I. that I want to slap someone every time they make a comment about the machines taking over. Only recently has the combined computing power of the entire world been enough to equal one human brain.

http://www.wired.com/wiredscience/2011/02/world-computer-data/

What Watson does is impressive but it is geared solely for that purpose. It does not have the methodology to think, let alone the hardware.

I honestly feel that Watson cheated by being deaf and blind, and having the clues emailed to him.

If they had wanted to make it a fair contest, they would have had to have made Watson do speech to text as Trebek read the clue and OCR the text version on the clue board, like human contestants have to.

Giving the clues to him the way they did is like telepathically beaming them directly into a contestant's brain.

>>2927
>What Watson does is impressive but it is geared solely for that purpose.

True enough. It's clear what Watson can parse natural language to a degree that gives it a (usually) high probability of guessing the correct question for an answer. But beyond the attempt of understanding the language we are left with a relatively simple to understand computer system.

So let's get down to the old question: What defines Artificial Intelligence?

>>2928
Yeah voice interpretation tech actually needs a lot more work in the realm of simply making the computer understand natural language first. Hence Watson.

>>2929
I'm sure there's a good dictionary definition somewhere but essentially it's Autonomous Thought; output generated by the computer without outside input.

>output generated by the computer without outside input.

I'm tired and hopped up on meds right now, but is that even a fair assessment? Look, even as babies we can't learn anything without responding to outside stimuli.

>>2931
welcome to the issue. I didn't really word it to well but in abstract we're talking about the machine being able to think, reason and create on its' own with no prompting from a user and no external prerogative that it is following for another entity. Human children are constantly surround by stimuli, yes, but built within each human brain is the ability create logical constructs regardless of outward stimuli. Granted one could argue that a human brain is never without outward stimuli, but we still can't make a computer that is as fast or as complex as the human brain, so the point's academic atm.

I think the machine is little different from being to English class what the calculator is to Math class. An unthinking, unfeeling, unemoting bank of data no different from a Wiki hooked up to a see&say.

That's not intelligent. That's a striped down ricer of a computer geared to do something complex and requiring a great deal of information to sift through to do it. Language at its core is a vexing, irritating and ambiguous thing that we use to communicate and convey thoughts to eachother. But it's still just data and the structures it's made of. It's as close to a consensus to different things as we can get, because we all see everything up to different shades of blue and green slightly differently. Even feelings, pain tolerance and sense of taste. But it still has a mathematical end of the road that can be programmed into a machine.

Is it an accomplishment for dealing with human speech? Absolutely. Positively. Undoubtedly. Yes.

Is it a sign of artificial intelligence?... We don't even have a clear and concise definition of intelligence yet, because we're still choking over our own egos over what makes us smart, sapient and alive. The way I personally define intelligence is the ability to take data and run it through a dynamic process to put it to use and then at least formulate, if not initiate, the process to do it. In nature, we developed this intelligence to see the apple, climb the tree and eat the fruit. We were driven to this smart move by our ancestors first learning to see the apples, climb the tree, and eat the fruit.

A computer is not the product of millions of years of reproductively selective evolution. It has no instincts or debatable instincts handed down over millennia. No wayward neurochemistry. It doesn't dream. It has no innate drive to survive or use new data in the quest to find food, shelter, fuck and then die. It has no reason for even existing or caring or feeling outside of what it's programmed to simulate doing. We're still just now working on the how and why to emulate all these individual functions of a working biological being's brain.

If you define intelligence as an organism capable of making some small sense of the world independent of another being asking them why and telling them how, no. Watson is not intelligent. If you define intelligence as the quality that makes one person able to solve math problems and retain information at school, then there are notebooks, calculators and chalkboards more intelligent than some human beings are on this world today.

I'd liked to see them do one game with Watson connected to the net just searching by itself without search engines. Would have been interesting.

>>2935
I'd be interested to see it as a search engine. Adapt the code to search the internet. "Semantic Web" Indeed.

>>2933
This would be the problem if we could actually build a single machine with the processing power of a human brain; How does one really define "Intelligence"? Moreover how do you define instinct for a lifeform that has never really had to fend for itself?

>>2936
>How does one really define "Intelligence"?
Denuded of philosophy and brought down to brass tacks, I think it might be the capacity to learn something and develop concepts and thoughts that the organism, or unit, wasn't programmed to do. An organism like the Pistol Shrimp isn't intelligence when it shoots prey with a vaporized bubble as hot as the surface of the sun.. an organism that figures everything out about how that words from the ground up and has the capacity to build dynamite, would be.
I dunno how the science world will define it, but like most everything else, I'd trust a materials scientist, mathematician and computer scientist to define intelligence before I'd ask a philosopher. Philosophers are the Dr. Phil of hard sciences. At least when the hard science folks don't know the answer, they won't gloss it over with philosophy.

>Moreover how do you define instinct for a lifeform that has never really had to fend for itself?

Well.. that's the funny thing. I think in order to properly answer this question, the values and perspective we should adopt would be similar to the way we changed perspective away from the bible and more towards impartial, neutral territory. Instinct is a biological function that developed how it did, because it could. It's not the best, it's not the worst, it's just the one that happened to be functional enough to carry down over generations.
What is instinct? Really it's just knowledge and information without a learned source, communicated over generational lines. Wushu you inherit from your family and never need to practice to know it. Because your folks and kind use it just to survive and function in the environment you exist in.
With this in mind, I'd say the closest parallel to that for computers removed from reproduction and natural selection would be Drives and Instruction Sets combined with hardware geared towards particular performances, and designated function. Series/models. It's not a perfect meeting to the definition, because the way we define instinct, intelligence and sapience is anthropocentrist.

I suppose then that a properly anthropocentric compatible definition of computer sapience and intelligence would be a self-actualizing robot that could actually feel and relate, capable of receiving and interpreting sensory data and measuring how the data affects their selves, and what it means to them. The ability to learn from these experiences and grow. But how a lack of real life/death or something as fickle as neurochemistry would affect this development, I have no idea.

To OP: No and No, but Watson represents a small, albeit significant advancement. You don't get to be the dominant species on a planet, capable of wiping out all life and landing men on the moon, without first learning to make fire and develop agriculture.

My personal prediction is that we won't be able to have true AI until we've surpassed the limits of transistors and conventional microchips. Quantum computing, protein and DNA-based computing, artificial neural networks, and combined read-write processor/data storage devices will be the next big things, and be as big an advancement as going from vacuum tubes and punch cards to Watson overnight. And, of course, computers with the data storage, processing power, and compact size of the human brain will also be a prerequisite for any future brain-uploading.

Until then, robots and specialized computers like Watson will continue to advance, mastering more and more complex tasks (first chess and composing music, today Jeopardy, tomorrow...?). We might soon get to the point where we can have computers who, while not true AI, will be able to outperform humans, in both speed and accuracy, in not just one extremely specialized task, but in a dozen or so related tasks, with minimal human input. Data processing, secretarial and administrative work could all be performed by such devices.

But OP is right about Watson missing context. Humans think in words, symbols, images, lofty concepts and basic, animal desires. The feedback loops in our neural networks are so perfectly ordered and yet seemingly random enough to allow for inspiration, fantasy, strokes of genius, induction, deduction, and guessing, both educated and shot-in-the-dark. Until we have a computer that actually thinks-- talking to itself and thinking in full sentences and recalled images--it'll be nothing more than machine.

I really feel this whole thing was silly. In all honesty, all three contestants (and myself, at home) knew the answers to the vast majority of the questions. The ONLY reason Watson won was not because he had a bigger base of knowledge than the humans, (I bet my balls its not as much by HALF, if that) but because it was capable of buzzing in faster. Only in the last few minutes of the last episode did the two humans start to be able to catch up. You'll notice Brad especially began to beat Watson's speed late in the game, but it was too little too late. Give them a rematch, and I think they can win.

This made it obvious that Ken Jennings had gotten rusty. After all, the biggest reason HE went on his 74-episode winning streak was the fact that he got SO fast with the buzzer, other contestants were unable to keep up. They actually had practice-rounds before the show to let them get familiar with the timing of buzzing in, to give them half a goddamn chance, and it still did not work.

This reminds me of being back in high school on Academic Team. There was one school that won first place almost every year. They also used different buzzers! The ones they used were shaped like sewing-machine pedals, instead of the thumb-triggers everyone else had. Nobody seemed to believe me when I said that gave them an unfair advantage when playing home games. Fucking idiots, the lot of them. At least in AT, though, you can ring in an answer whenever you want; no waiting until the split-second the host stops talking.

So yes, I wasn't impressed with this at all. Because Jeopardy is 20% a game of knowledge, and 80% a game of speed. Ken Jennings proved that. Machines are always FASTER, but it's going to be a very long time before they are SMARTER. Watson was a damn good search engine, better than any other, but it's frankly not intelligent at all. At least how I would define it.

Just wanted to say that.

Great points. Maybe they need to eliminate the speed and do nothing but final jeopardy type rounds for a more accurate competiton?