Weak AI beat Chess, now on to Jeopoardy!

[ElectricFetus]

http://www.youtube.com/watch?v=qO1i7-Qx00k&feature=fvwrel

IBM's Watson AI program can analysis jeopardy questions (don't know if it gets the questions verbally, visually or as text, but it must know when the questioner ends the question to "hit the button") and can process a correct answer in seconds, fast enough to compete with the worlds best jeopardy champions.

 

[EmptyForceOfChi]

Thats cheating lol.

It's like saying how good a Calculator is at Mathematics. Or how good a kettle is at boiling water.


It's all relays, self analysis and processing proggrammed hardware with software. The machine will never have consciousness and ability to truly learn and understand what it is learning.

I do enjoy playing vs chess AI's but they are still cheaters.

Peace.

 

[Michael]

I OTOH think "conscious" AI will arrive sometime in our lifetime. One day AI may be more "conscious" then humans. There's no reason why not. There's nothing special about humans that preclude any other groupings of atoms to attain consciousness.

 

[domesticated om]

This has "new search engine" written all over it.

Watch out google

 

[Sarkus]

Thats cheating lol.

How so? It can only use the information stored in its data-banks (which it has collated over the past 4 years or so) - just as a human can only use info stored in their brain.

The "AI" aspect of it is in the interpretation of the question... how well can a machine understand complex and enigmatic questions, and then pick out a precise answer from its memory.

Apparently in trials it wasn't too bad, although probably not up to the speed of the best Jeopardy contestants that it will face on tv.

It's like saying how good a Calculator is at Mathematics. Or how good a kettle is at boiling water.

Or how good a person is at interpreting questions and coming up with the right answer?

It's all relays, self analysis and processing proggrammed hardware with software.

Are you talking of machines or humans here??

The machine will never have consciousness and ability to truly learn and understand what it is learning.

The latter could easily be said of the majority of humans.

Also, what do you think it is to "understand" something?

 

[Read-Only]

I OTOH think "conscious" AI will arrive sometime in our lifetime. One day AI may be more "conscious" then humans. There's no reason why not. There's nothing special about humans that preclude any other groupings of atoms to attain consciousness.

Some people put too much blind trust into what technology can do - and this is a good example of that. We cannot even figure out what consciousness IS - much less create it in a machine. :shrug:

 

[ElectricFetus]

It's all relays, self analysis and processing proggrammed hardware with software.

Perhaps we are all this as well and consciousness is an illusion?

The machine will never have consciousness and ability to truly learn and understand what it is learning.

Perhaps for a turing machine, but I think your are wrong about all machines, as we are machines, therefor machines have already achieved consciousness. Even if only a biological or analog machine is the only thing that can achieve consciousness eventually technology will make it, if technology continues to progress as it does.

Unless your saying we have properties that make us conscious that are beyond the physical realm and thus can never be replicated or emulated, I'm not sure this is true but I think it worth testing. If we ever do make a Strong AI that is as smart or smarter than any human in every way and that claims to be conscious we must accept these conclusions:
1) Conscious can be an illusion, as the AI thinks it conscious but we disagree, ergo we may also be delusional about are own consciousness.
2) Consciousness is completely physical and can be replicated with known machinery (as opposed to the present state of biology which has unknowns) and thus a soul is irrelevant.

Certainly these are answer that could radically change our philosophy and theology, scary answers in fact, but I'm going to wait and see if these answers are true or not before passing judgment on them.

I do enjoy playing vs chess AI's but they are still cheaters.

It cheating only if consciousness is required to play a game.

Some people put too much blind trust into what technology can do - and this is a good example of that. We cannot even figure out what consciousness IS - much less create it in a machine. :shrug:

We don't need to know what it is, we could still replicate it via a black box system, many evolving programs for example already derive answers from which the creators are baffled at how it came to those answers. We don't know if consciousness is an emergent property that a machine sophisticated enough might not simply become self aware.

 

[cosmictraveler]

http://www.youtube.com/watch?v=qO1i7-Qx00k&feature=fvwrel

IBM's Watson AI program can analysis jeopardy questions (don't know if it gets the questions verbally, visually or as text, but it must know when the questioner ends the question to "hit the button") and can process a correct answer in seconds, fast enough to compete with the worlds best jeopardy champions.

It recieves the questions in text format.

It did make mistakes and it couldn't hear the other contestants wrong answer and answered the wrong way as well.

 

[ElectricFetus]

It recieves the questions in text format.

It did make mistakes and it couldn't hear the other contestants wrong answer and answered the wrong way as well.

I never said it was perfect, just that it could compete with them. I don't think its possible to make an omnipotent machine, we would need infinite processing power (possible assuming temporal computing) and all the information in the universe (most likely impossible)

 

[cosmictraveler]

I never said it was perfect, just that it could compete with them. I don't think its possible to make an omnipotent machine, we would need infinite processing power (possible assuming temporal computing) and all the information in the universe (most likely impossible)

One day they could tie their machine into the WWW and then it would be even more advanced. Right now Watson is only using its internal programming to answer these questions.

 

[ElectricFetus]

One day they could tie their machine into the WWW and then it would be even more advanced. Right now Watson is only using its internal programming to answer these questions.

Not unless the Web got much much faster, to derive any answer it would need to download a huge amount of information simultaneously.

Here how Watson was built.
http://www.youtube.com/watch?v=3G2H3DZ8rNc&feature=relmfu

 

[fedr808]

all this means is that it is a very advanced, very effective search engine.

 

[adoucette]

Well, it's internal programming includes data from more than one Million books. But the key, to playing the game of Jeopardy is that you need to have only trusted sources of information.
Thus adding the WWW wouldn't necessarily make it better at the game because of all the BS on the Web.

Arthur

 

[Read-Only]

We don't need to know what it is, we could still replicate it via a black box system, many evolving programs for example already derive answers from which the creators are baffled at how it came to those answers. We don't know if consciousness is an emergent property that a machine sophisticated enough might not simply become self aware.

Sorry, but you've seriously shot yourself in the foot with THIS one! Please show proof of "...already derive answers from which the creators are baffled at how it came to those answers" - I think that's nothing more than an urban legend or false thinking on your part.

 

[ElectricFetus]

Sorry, but you've seriously shot yourself in the foot with THIS one! Please show proof of "...already derive answers from which the creators are baffled at how it came to those answers" - I think that's nothing more than an urban legend or false thinking on your part.

First of all, are you saying that we can't make things that we don't understand?

 

[Read-Only]

First of all, are you saying that we can't make things that we don't understand?

Nope, no changing the topic - answer the question, please. If you can.

 

[Stryder]

Some people put too much blind trust into what technology can do - and this is a good example of that. We cannot even figure out what consciousness IS - much less create it in a machine. :shrug:

Consciousness is just a cycled pattern of reiteration through a biological logicgate network. a Biological system's logicgate's however can alter in shape and form to either support or disrupt continuation. It's the main reason why if a person gets a knock on the head they can suffer amnesia and why when you sleep at night you are unaware of the passage of time etc, as each of those has instances where areas of the brain are no longer accessed or are suffering from a disruption.

Death occurs naturally when the Conscious cycle can no longer be reiterated due to cellular ageing.

I guess you could say that Consciousness is just continued re-evaluation of property states while it is still possible to evaluate.

 

[ElectricFetus]

Nope, no changing the topic - answer the question, please. If you can.

Ah, but can I? Certainly I could spend hours mining for a quote, but I don't have the time or desire to do so, does that mean I'm wrong? Does lack of evidence for something prove its not there? So to answer your question, No I can't, but that does really answer your question, sorry. :shrug:

But certainly if you wish to believe genetic algorithms have neither surprised or even dumbfounded a single one of their creators I'm not stopping you. I'm just trying to contextualize that belief into something relevant to this thread.

 

[fedr808]

Consciousness is just a cycled pattern of reiteration through a biological logicgate network. a Biological system's logicgate's however can alter in shape and form to either support or disrupt continuation. It's the main reason why if a person gets a knock on the head they can suffer amnesia and why when you sleep at night you are unaware of the passage of time etc, as each of those has instances where areas of the brain are no longer accessed or are suffering from a disruption.

Death occurs naturally when the Conscious cycle can no longer be reiterated due to cellular ageing.

I guess you could say that Consciousness is just continued re-evaluation of property states while it is still possible to evaluate.

I disagree, death can come from numerous factors, how many fatal diseases are the direct result of neurons aging?

Sure there are a lot, but by no means are they the majority of deaths.

And your argument has a couple of flaws. First off, what governs what these 'logicgates' consider to be supportive or disruptive?

Where does emotions fit in?

Very little in life is black and white, yes or no, how would this hypothesis of yours consider this fact?

Where does the subconsciouss fall in?

 

[ElectricFetus]

Well I would suppose emotion, subconscious, etc, are all just outputs of the neural network, we have no evidence that it comes from anywhere else, and through manipulation of the brain we can alter emotion, personality and memory, so all of these functions seem to be there in the physical brain, and thus if its physical its functions can be replicate, emulated, simulated, either by direct physical copy or indirect simulation, unless there really is something supernatural to the mind that can't be detected or measured.