Could Deep Blue play a smarter AI ?

[5cats]

Good, good... Does He also talk to you when you don't forget to take your meds ?



If I remember correctly, the arrow paradox says that "to reach a target, an arrow first needs to cross half of the distance to the target, then half of the remaining distance, then half of the remaining distance, etc ad infinitum. So in effect the arrow can never reach the target." That sounds to me like a classic case of infinite sum giving finite result. In this version at least, I don't really see what time's got to do with it (unless you say it will take an infinite time for the arrow to reach the target, but that's the same thing).

Tsk! He wasn't talking to me, I was reading a book (in my dream) and God was talking to a dwarf robot who'd developed a soul. Now my cats, they talk to me aaaall the time...

I'll try to find a more complete example of Zeno's Arrow, but basically it postulates that, if at any give POINT in time, the arrow is not moving, how then can it continue motion between the different points? (Aristotle had figured time was made of indivisible "points" which bugged Zeno) There's a little more to it, but as I said, it's a paradox about the nature of time.

Here's a couple, not as complete as I'd hoped but better than Wiki.

http://www.geocities.com/capitolhill/lobby/3022/arrow.html
http://faculty.washington.edu/smcohen/320/ZenoArrow.html

 

[dbergan]

Gee, I wouldn't want to see the others in the class... Just kidding. But I would say though, David, that although I would tend to be more on "your line" of "everything not being reductible to matter" (although I think I'm much less dualistic than what you seem to be), your insufficiently rigorous grasping (or maybe exposition) of logic and the founding principles of modern science tends to weaken your stance and make you an easy target for the guys "on the other side". That's too bad. However, I salute the passion with which you defend your point of view. And we know that, in history, it's more about "who wants it the most" than about "who is the rightest"...

Thanks... I think. Care to elaborate on "your insufficiently rigorous grasping (or maybe exposition) of logic and the founding principles of modern science"? If I'm missing something, I would truly like to learn.

 

[dbergan]

No it's not. Religion can borrow elememts to philosophy, but it's first and foremost a BELIEF system. You don't even need to understand (although most theologians are also brilliant philosophers, see Saint Augustine or Saint Thomas Aquinus), you just need to believe. In religion, the Truth comes from Revelation, not logical thinking (although logical thinking can be used afterwards to try and justify the Revelation, see the two above mentioned theologians for instance and even some posts in this threads (see what geniuses we are ?) ). Philosophy, in its westerm fom, is the quest for wisdom (philosophy means "love of wisdom" literally) through reason. Philosophy does not claim to have the answers for everything like most religions do.

I don't totally agree, nor do I totally disagree. Just keep in mind that in addition to what you have said, there is a very real element that religions (especially in primitive places/times) are based on the concept that there are spirits here and there. There are light and dark elements to religions... the light is the rational theological parts. The dark is the spooky mysterious stuff. We focus so much on the light aspect of religion that we think that people must have been mad to think that spirits/demons/witches existed and had influence... yet it is undeniable that religion was born from this rather than philosophy. Certainly you can say that we don't believe in it, but we cannot say that they didn't believe in it. And from what I hear from current living missionaries, if you go to Brazil, Haiti, or parts of deep Africa, you will find convincing proof of witchcraft and voodoo.

You don't have to believe it, but don't deny it unless you've been there.

Then you're being sophistic... The question is not : "can it be done while still being a homo sapiens" (it takes a 's' even in the singular, it's latin meaning "wise"), because frankly who cares, this protoplasmic form known as "homo sapiens" derived from another form, and will (is) most certainly evolve (evolving) into yet another form, either by natural or artificial means. The question is : "can it be done ?", and if you give it enough time and motivation, the answer has a very good chance to be yes. See what we, collectively as mankind, have accomplished in only about 10000 years of history. And you probably can bet that what seems to be the intangible limitations of today will be broken tomorrow. Even lightspeed limit, for instance, might be "walked around" by using wormholes, or mastering the hidden dimensions postulated by the Superstring theory (just making wild guesses here...)

I can't believe people are falling for this "stay open-minded, with science anything is possible" idea. Take a video-tape of a car accident. Then watch it in rewind. Or video a cup of coffee cooling to its surroundings... then watch it in rewind.

Or consider the fact that matter/energy cannot be created or destroyed.

Or that it won't snow on a sunny day when its 100 degrees outside.

Science establishes the limits of what is possible and impossible (by natural means), it doesn't make everything possible.

 

[5cats]

^^ true! Some things, by definition or by cruel reality, are simply not possible.

Sing along everyone!

"Heat is work and work is heat (repeat)
You can't take heat from the cooler to the hotter (repeat)
You can try it if you like but I don't think that you oughter..."

Love that song!

 

[Brighteye]

If Firaxis was able to create the perfect Civ AI - namely AI that plays better than 99.99% of humans and has the ability to adapt.. this would then become the hardest difficulty level.

It would then be fairly easy to dumb down the AI to accomodate the easier difficulty levels. No handicaps would be required.

Nice! Can you try to give a simple (for a guy who's not a computer scientist) explanation of how to make a perfect AI more stupid, rather than simply introducing random errors? Or were you just saying that if we can do something as hard as make a perfect AI then we should be able to crack the lesser problem of making it stupid, even though 'we' don't know how at the moment?

 

[Brighteye]

I don't know about other people, but personally that's exactly the kind of AI i'd like to face, and the kind of situation i'd love to have to deal with. For one thing, competing against myself is much more rewarding than competing against a computer. For another, a perfectionist like myself would love a game where the whole point was to make as few mistakes as possible.

I'd have thought that most people play a game in order to compete at a more abstract level, so that rather than comparing ability to micromanage and be perfect, they want to devise strategies that are more cunning than their opponents'. Thus an AI that could foresee all outcomes and always countered your strategies, but could be beaten by taking best advantage of your bonuses, is not fun.
I thought that people want to play strategy games for the purpose of testing the success of their strategies, whereas a perfect AI would always win on this count, but only be beaten by greater resources.
Thus there are two different factors in the game; the strategies after which it is named (Turn-based strategy) and the management, which adds some fun in as far as it allows for greater variations in strategies.
However, if lots of people are more like you, and enjoy chasing perfect micromanagement more than devising strategy then the task of making a much-appreciated AI is easier, because instead of requiring intelligence, all that is required is any old way of making the game difficult.
I personally want the AI to mimic intelligence as closely as possible, in that I see intelligence as the only means to devise strategies, and it's the conflict of strategies (and the eventual triumph of mine) that makes a game fun. An AI with no strategy but perfect micromanagement is boring. An AI that's been programmed to have stupid strategies, which are made up for with AI bonuses is even less fun. I want my strategies to be competing with other strategies, not bonuses.

 

[Brighteye]

No it's not. Religion can borrow elememts to philosophy, but it's first and foremost a BELIEF system. You don't even need to understand (although most theologians are also brilliant philosophers, see Saint Augustine or Saint Thomas Aquinus), you just need to believe. In religion, the Truth comes from Revelation, not logical thinking (although logical thinking can be used afterwards to try and justify the Revelation, see the two above mentioned theologians for instance and even some posts in this threads (see what geniuses we are ?) ). Philosophy, in its westerm fom, is the quest for wisdom (philosophy means "love of wisdom" literally) through reason. Philosophy does not claim to have the answers for everything like most religions do.

Yeah, that's weird how a conversation about logic can turn so personal... Guess now we can understand why people back then would burn each other up (litterally) over such arguments when they had the possibility to do so...


That's because you presuppose that there isn't any "hidden ingredient" in the making of a human being (or any lifeform for that matter) other that the chemical transformations we can witness. And if this is the case, nothing says this "hidden ingredient" can be mastered other than through natural processes. Yes, Nature can and does manufacture living beings. That does not mean we can replicate it. However, in time, by developing and mastering some of the AI techniques described in this thread and by inventing others (which will inevitably imply some form of learning), I don't see why we wouldn't be able to manufacture a machine which perfectly emulates sentience (yes, emulate sentience, not being sentient, there IS a difference, but again it's metaphysics...)





Then you're being sophistic... The question is not : "can it be done while still being a homo sapiens" (it takes a 's' even in the singular, it's latin meaning "wise"), because frankly who cares, this protoplasmic form known as "homo sapiens" derived from another form, and will (is) most certainly evolve (evolving) into yet another form, either by natural or artificial means. The question is : "can it be done ?", and if you give it enough time and motivation, the answer has a very good chance to be yes. See what we, collectively as mankind, have accomplished in only about 10000 years of history. And you probably can bet that what seems to be the intangible limitations of today will be broken tomorrow. Even lightspeed limit, for instance, might be "walked around" by using wormholes, or mastering the hidden dimensions postulated by the Superstring theory (just making wild guesses here...)

Here again we have misinterpretation of words. Religion is a philosophy, because 'a philosophy' is a system of thought. Religion is not philosophy, because 'philosophy' (without the indefinite article) is a way of trying to discover truth/wisdom.

All debates very easily turn personal. If you destroy someone else's argument you are, deliberately or entirely innocently, showing them up as more stupid than you are. At least, it's very easy for someone to feel as if you are. That's almost a personal insult... and so we have flamewars. It's harder to have a truly friendly, warm, lovey debate than it is to devise the cunning arguments that form part of the debate.

We're dealing with a scientific problem, of creating intelligence. We can see a brain growing, we know some of how it develops. Eventually, we may be able to imitate this precisely. This imitation must function as the original does. If it does not, then we must rewrite science and our understanding of the world. Thus, if you accept the principles of science (and consequently, share our current understanding of the world as defined by science) then you must believe that there is the theoretical possibility of creating AI.
If you postulate some hidden ingredient, how does it get there? Why would it not go into our AI? An important part of science is taking the simpler theory. If you won't do that, why accept any of the principles of science? At the moment the theory of an invisible soul giving awareness is as sensible as an unknown process in the complex organ that is the brain. However, if we create a similarly complex imitation then it should have the same awareness, whether it be through a soul or a mechanistic process.

Maybe his example was bad, but the general statement that 'Some things are impossible' is a true one. We can prove that some things are impossible; a mathematician should be able to recall some sort of proof that something can't be proven, although a suitable example to demonstrate the statement eludes me.

Can someone tell me how to do multiple quotes?

 

[White Elk]

Can someone tell me how to do multiple quotes?

Based on your inabilty to discover how to post multiple quotes.. Would you have thought it impossible if you had not seen others do it? What if people had also told you that it was impossibile.. would you have believed them? Would you have then passed on this erroneous info to the next guy?

There are no posting controls for multiple quotes, so you will have to do it manually. So start a new post and then scroll down to the "Topic Review". Locate the text you wish to quote, highlight and copy it. Then scroll back up and paste the text into the posting window. Then highlight the text again and press the

icon, located just above the posting window. But this will result in a generic quote that is not attributed to anybody. In order to identify the original poster you will have to modify the quote tag by addiing "=PostersName" directly after the first Quote tag, but within the brackets. examples...

[ QUOTE ] Blah Blah Blah... [ /QUOTE ]

[ QUOTE=Brighteye ]...Yackity Smackity[ /QUOTE ]

There are other ways to go about this, but this will get you started. If you ever wonder about how somebody did something.. then just quote their post and investigate. You will see what tags they used and then you to will have the knowledge.

EDIT: I just realized that quoting a post will result in the loss of the quotes that were contained within the original post. So this quote/investigate/learn method does not work in every situation.

 

[Brighteye]

Thanks. I wondered if I'd have to write out

every time, but this sounds easier.

 

[White Elk]

Also.. if your quoting the same poster multiple times, you only have to write it out once and then you can copy/paste the other quote tags. If I've alot of quotes to post then I find it expediant to copy/paste the quote tags ahead of time. Then I add the quote between the tags as I go. Like so..

[ QUOTE=UnderDog ] [ /QUOTE ]

[ QUOTE=UnderDog ] [ /QUOTE ]

[ QUOTE=UnderDog ] [ /QUOTE ]

[ QUOTE=UnderDog ] [ /QUOTE ]

etc...

 

[jar2574]

While it may be theoretically possible, AI based upon brain replication remains well beyond our reach.

We can see a brain growing, we know some of how it develops.

We cannot see a brain in a fetus during the first trimester, unless we take the fetus outside of the mother, at which point brain development promptly ceases. This makes exact imitation of growth and development very challenging for something as complex as the brain, with its billions of neural connections.

Eventually, we may be able to imitate this precisely. This imitation must function as the original does.

Even if we create an exact imitation of the brain, the imitation will not perform as the original does unless we can surround the brain with a real or a replicated human body. Otherwise the imitation will not function as the original does.

Unless ethical/legal standards change prior to this scientific development, the legal hurdles around research where an imitation brain is placed in a original or replicated body would seem to be enormous. (But if we can replicate a brain then a few legal hurdles might be the least of our worries.)

 

[korsteve]

This thread may have gone off in a different direction but in regards to the OP I think that chess programs are so strong because it is relatively easy to evaluate a board position. This allows the program to choose the best out of a large number of possible moves, predict opponent responses, counter, etc.

Also another important factor that I didn't see mentioned is that the game starts with all the pieces on the board and there are no real surprises. There are tight constraints on possible moves and there are no unknown factors. With games like Civ and Go the environment and board are always different. Because of this, the AI for Civ is probably mostly based on Heuristics while chess is probably solved using some kind of global optimization algorithm.

 

[Zombie69]

It's the electrical relationships of atoms that determine what is what and how things react to each other.

There are 4 basic forces governing atoms and everything else in the universe. Electricity is not one of them. The forces are electromagnetism, gravity, strong nuclear force, and weak nuclear force. All of them affect atoms, though gravity has very little effect because of the small masses involved.

And the electric impulses in the brain are much too weak to have any noticeable direct impact on outside bodies.

 

[Brighteye]

While it may be theoretically possible, AI based upon brain replication remains well beyond our reach.

We cannot see a brain in a fetus during the first trimester, unless we take the fetus outside of the mother, at which point brain development promptly ceases. This makes exact imitation of growth and development very challenging for something as complex as the brain, with its billions of neural connections.

Even if we create an exact imitation of the brain, the imitation will not perform as the original does unless we can surround the brain with a real or a replicated human body. Otherwise the imitation will not function as the original does.

We don't need a full body; just a system of inputs into the nerves. That'll probably the least of our problems by the time we can recreate a brain.
True, we can't really see a whole brain growing, but we can do experiments on nerve migration/development and do dissections at different times during development.

And the electric impulses in the brain are much too weak to have any noticeable direct impact on outside bodies.

Absolutely. Your brain impulses are tiny. I've done experiments measuring them, and the subject has to stay very still because any jaw or even eye movement overpowers the weak nerve signals with the muscle impulses. The amount of voltage generated by a large muscle is vastly greater than anything the brain comes up with.

 

[White Elk]

There are 4 basic forces governing atoms and everything else in the universe. Electricity is not one of them. The forces are electromagnetism, gravity, strong nuclear force, and weak nuclear force. All of them affect atoms, though gravity has very little effect because of the small masses involved.

Huh!?! ~ Electrons in motion are what constitute electric current in the first place.


Electromagnetism
1. Magnetism produced by electric charge in motion.
2. The physics of electricity and magnetism.

Electricity
1. The physical phenomena arising from the behavior of electrons and protons that is caused by the attraction of particles with opposite charges and the repulsion of particles with the same charge.

And the electric impulses in the brain are much too weak to have any noticeable direct impact on outside bodies.

Atoms Can gain and lose electrons with ease. And it is the sharing of electrons between atoms that determine what a thing is and what state it is in. Simply by touching ice, we create an atomic reaction. Even the water vapor we release through breathing creates atomic reactions in our enviroment. Imagine a line of people standing side by side and shoulder to shoulder. If you were to push one persons shoulder, then they would all move. Newtons Third Law of Motion states that.. "For every action there is an equal and opposite reaction." This applies to not only contact interactions but also distance interactions caused by gravitational, electrical, and magnetic forces.

So, since electrons are in a constant state of motion and freely interact with the electrons from other atoms.. and the fact that a change in one atom will effect all the atoms around it creating a domino effect.. I say why wouldn't it be in the realm of possibility that the human brain could possibly effect change on the atomic level?

 

[White Elk]

And the electric impulses in the brain are much too weak to have any noticeable direct impact on outside bodies.

Absolutely. Your brain impulses are tiny. I've done experiments measuring them, and the subject has to stay very still because any jaw or even eye movement overpowers the weak nerve signals with the muscle impulses. The amount of voltage generated by a large muscle is vastly greater than anything the brain comes up with.

It's the brains tiny impulses that trigger the powerfull reaction of a muscle contraction in the first place. And that tiny impulse yeilds a powerfull reaction. Why wouldn't that effect be present in other interactions?

 

[NapoléonPremier]

You think i don't know all this already? Or maybe you were not talking to me directly but explaining my position to others?

Think of it this way. Newtonian mechanics were not proven wrong by Einstein's theory of relativity. They were merely brought down from a set of equations that were meant to explain everything all the time, to a set of equations that are true only for a specific set of conditions (mainly macrospopic objects traveling far below the speed of light). They're still true, but their domain of application has been restricted.

In the same way, i believe that quantum mechanics, and particularly the parts that introduce an element of randomness, will be refined in this way. I'm not saying that they're wrong per se; what i'm saying is that they're only valid in a specific set of conditions, namely when we don't have access to certain bits of information. And in cases where we do have access to this information (not accessible today mind you), another set of equations will yeild more accurate results, and hopefully this set of equations will include no random elements (else it will need to be further refined).

I didn't mean to belittle you. But it seems to me you have an almost religious belief in determinism. As has already been mentioned several times in this thread, quantum mechanics is not just uncertainty in the measurement or available information. For now, it seems that uncertainty is really a central elememt of the system. Saying otherwise without new elements is just a religious belief, or a petition of principle.

 

[Zombie69]

Huh!?!Electromagnetism
1. Magnetism produced by electric charge in motion.
2. The physics of electricity and magnetism.

You've got electromagnetism all wrong. Pick up a book on electromagnetism and you'll understand what i mean. The second definition you give is good, but doesn't come close to showing all the implications.

For instance, for an electric current to act at a distance, you need to consider electromagnetism. The electric current creates a magnetic field, which in turn creates an electric current somewhere else. By the way, there's no such thing as an electric field.

So, since electrons are in a constant state of motion and freely interact with the electrons from other atoms.. and the fact that a change in one atom will effect all the atoms around it creating a domino effect.. I say why wouldn't it be in the realm of possibility that the human brain could possibly effect change on the atomic level?

Well, for one thing, axions are covered by a sheet that's an electric insulator. Good thing too, otherwise they couldn't carry currents properly. Also, your cranium isn't particularly conductive either! And neither is the air separating you from the object you intend to affect.

I think you seriously need to take a basic course on electromagnetism before you start making claims like these.

 

[Zombie69]

It's the brains tiny impulses that trigger the powerfull reaction of a muscle contraction in the first place. And that tiny impulse yeilds a powerfull reaction. Why wouldn't that effect be present in other interactions?

In order for the neuron to give the signal to the muscle, it must emit a molecule in the medium. That molecule reaches a receptor in the muscle, which starts a reaction that will cause it to contract. There's no electric connection between nerve and muscle. In fact, there's not even any electric connection between two neurons; they interact with transmitters and receptors in the same way.

These molecules that are emitted by neurons are not sent into the air! Even if they were, they wouldn't survive there, as they need a water based substrate. And even if they could survive, they would have no effect since they're just proteins that need the appropriate receptor on a cell in order to affect that cell. And if you're talking about inanimate objects, there's really zero chance of that happening since they don't even have cells to begin with.

 

[Zombie69]

For now, it seems that uncertainty is really a central elememt of the system.

Because we can find no cause for certain microscopic events. But that doesn't mean this cause isn't there. It just means we haven't found it yet. Indeed, it would be hard to see that A affects B when we don't even know that A exists. But we will one day.

Up until recently, we've always been able to find a cause to everything that we observe; there's no reason why this should change. As far as i can see, believing in determinism has nothing to do with religion. It has to do with observation, not just in the present or in one experiment, but of everything we've ever studied in the history of science. Kind of seeing the forest from the tree. And from that observation, making the most plausible observation. Making conclusions based on observation isn't religion, indeed this is precisely how science works.

Observation : up until recently, we've always been able to find a cause to everything that we observe.
Most plausible conclusion : the universe is deterministic, and anything for which we can't find a cause now isn't actually random, but rather caused by something we haven't figured out yet.