Could Deep Blue play a smarter AI ?

[Brighteye]

IMO there is somewhat of a middle ground. Causality does not govern quantum physics, but causality is not all wrong. It is just a theory (like all scientific theories) that only applies in certain circumstances.

I guess if your idea of causality is that causality must always be applicable to be correct, then yes, causality is all wrong under our current understanding of the world. I don't expect any scientific theory to be abblicable to everything, so I wouldn't call causality "all wrong." IMO we do not yet know how much quantum physics will change our understanding of other scientific laws, so we don't know just how "wrong" causality will end up being.


I understand where you're coming from. It appears that you think that science points so strongly to your conclusion that an assumption that your conclusion is true is warranted.

We disagree because in my view the most reasonable thing is to assume nothing, and to recognize the limits of science. We have not proven that intelligence comes from sources governed by the laws of causality. So I won't assume that it does.

Well, you and I may be reaching agreement, but I'm still discussing things with 5cats. I also have some things to say about this post. You missed what I said with the first two paragraphs quoted here. I gave two options:
1. Quantum effects have been shown to be outside of causality and therefore causality is all wrong
2. Quantum effects have been shown to be outside of causality and therefore causality is wrong with regards to quantum effects.
You mistook me for saying that causality either is all true or all untrue. My second option is not quite the same as all true. Causality is an idea that does not apply in all situations, but since we assume that it applies, as part of our understanding of the world, we require evidence that it does not before we ignore conclusions following from it.

If you think that the most reasonable thing to assume is nothing, then you do not assume causality at all, and you are a solipsist. I do not say that science points to my conclusion, but that the principles of science (and our understanding of the rules of the real world) are inconsistent with souls being relevant to making AI. The difference is that the principles are things we assume to be true before undertaking scientific experiment (or any action at all), whereas in your representation of what I say, my conclusion follows after the experiment. The experiment is not possible, so of course you'll think that I'm making it up.

 

[Dusty Monkey]

Perfectly!
Of course when an experiment is done, and it's results do NOT conform with "causality" it therefor MUST be based on an "unjustifiable premise"! Hang the results if they don't match YOUR theory!
Not "necessarily", but by our current understanding, YES!

You are attempting to use an experiment to prove a negative?



An experiments results need to be interpreted as to their ramifications. This interpretation can be logical or not logical. Obviously the ones that arent logical are bad interpretations.



I know of no experiments that disprove causality.. do you?

You must be relying on a logical OR illogical interpretation as to the ramifications of the results of some test (EPR paradox perhaps? Bells experiment?)

Since the experts do not agree with you that causality has been disproven, you should send them an email and we can end this debate now

There are multiple logical interpretations of quantum theory:

Copenhagen interpretation, Many-Worlds interpretation, The transactional interpretation, Everett's interpretation, Bohm's interpretation, and the Ithaca interpretation just to name a few!

Not all of these interpretations break causality. Some are indeed more favorable when Occams razor is applied but that doesnt apply to truth, only to presumption. You shouldn't be using presumptions to 'prove' things.

 

[Dusty Monkey]

Well, using Occam's Razor helps, let me elaborate:
The 'double slit' experiment defies all our current understanding. All the 'traditional explainations' we come up with are both inconsistant and/or extremely complex. (alternate realities, time travel, electrons having free will... ) SO we conclude that there's something else governing it that we currently don't understand.

Actualy we understand the double-slit results quite well. Quantum Electrodynamics (a subset of quantum theory) is heralded as one of the best and seemingly most accurate theories mankind has ever considered. The theory itself predicts some fundamental constants of the universe to a great deal of accuracy (our current inability to measure more accurately prevents further confirmation)

Don't believe me? Listen to (and watch!) Richard Feynman give a lecture on the subject of Quantum Electrodynamics at the Vega Science Trust:

http://www.vega.org.uk/video/subseries/8

I seem to remember than unfortunately it requires RealPlayer, but many people here should have that media player. I believe the entire series was around 3 hours long... BUT WELL WELL WORTH IT!


Feynman himself was considered to be one of the greatest minds AND educators of our time. You really need to see this video and there is a very good chance it will shatter your CURRENT view of what 'nondeterministic' means (not to say that a more mature view of nondeterminism doesnt exist.. one indeed does)

An example of how deterministic the double-slit results are can be found on many credit cards, childrens trading cards, many national currencies, as well as stickers --> The old favorite of marketing --> Holograms!

 

[jar2574]

If you think that the most reasonable thing to assume is nothing, then you do not assume causality at all, and you are a solipsist.

I am not a solipsist, but my remarks were probably unclear. When I said we should assume nothing, I meant that I don't think we should assume hypotheses are true unless they have been tested. If something is untestable and unproven, I'm comfortable not assuming a conclusion about it.

I assume that I will hit the ground if I jump off a building, because that's been tested. I don't assume that identical brains will behave identically, or that science can make AI identical to human intelligence, because those things have not yet been tested or proven.

I do not say that science points to my conclusion, but that the principles of science (and our understanding of the rules of the real world) are inconsistent with souls being relevant to making AI.

We'll agree to disagree I think.

The existence of souls is untestable. Therefore principles of science cannot be used to tell us whether souls are relevant to making AI or anything else.

Principles of science are not inconsistent with the existence of souls. Souls exist or don't exist outside the realm of science, because their existence cannot be scientifically tested. Science doesn't tell us anything about souls. Science can't prove that invisible monkeys don't pull some apples off trees in exactly the same manner as gravity pulls others.

Because no one has proven that intelligence only comes from physical sources governed by causality, the theory of causality is not inconsistent with the idea that souls may create intelligence.

The difference is that the principles are things we assume to be true before undertaking scientific experiment (or any action at all), whereas in your representation of what I say, my conclusion follows after the experiment. The experiment is not possible, so of course you'll think that I'm making it up.

I assume causality applies in my daily life and in scientific experiments. I don't assume causality applies to souls. And I don't assume that causality explains intelligence because no one has proven that intelligence can only come from physical things governed by causality or proven that human intelligence can be replicated by humans.

Personally, when we start making AI identical to human intelligence then I will assume that your conclusion is correct, even though "universal doubt" would still suggest that it could be false.

 

[atreas]

The classical text on AI is (currently) the book "Artificial Intelligence - A Modern Approach", by Russell and Norvig. The problem you are examining here is examined briefly in Ch. 26 of this book, in the section "Can Machines Really Think?" and the subsections "The mind-body problem" and 3 more subsections with thought experiments. Unfortunately my English are very poor to discuss this subject in full depth (I think everybody interested on the subject should have a look at the book), but the main points presented there are the following:

The basic thing that we have to decide is whether there are two distinct types of things (body and soul), which is called the dualist theory, or there is no such a thing like an immaterial soul, which is called the monist theory (more often it is called materialism).

Under the monist theory the mental states are just states of the brain. But let us examine a particular kind of mental state: the propositional attitudes or intentional states: like the state "I desire something", or more specifically "I desire an ice cream" or "I desire a pizza". Are these two states the same? The problem is that they refer to objects of the OUTSIDE world; hence, if we say they are different because they are connected to different objects of the outside world we violate the principle we stated before: that mental states are states of the mind (i.e. internal).

To examine this problem the AI scientists have proposed some thought experiments, which include:

The "brain in a vat" experiment
The brain prosthesis experiment
The Chinese room

You certainly need to have a look for these in the web to understand how present AI science views this problem (sorry I can't help more - as things get deeper my English fail me).

 

[5cats]

Actualy we understand the double-slit results quite well. Quantum Electrodynamics (a subset of quantum theory) is heralded as one of the best and seemingly most accurate theories mankind has ever considered. The theory itself predicts some fundamental constants of the universe to a great deal of accuracy (our current inability to measure more accurately prevents further confirmation)

Don't believe me? Listen to (and watch!) Richard Feynman give a lecture on the subject of Quantum Electrodynamics at the Vega Science Trust:

http://www.vega.org.uk/video/subseries/8

I seem to remember than unfortunately it requires RealPlayer, but many people here should have that media player. I believe the entire series was around 3 hours long... BUT WELL WELL WORTH IT!


Feynman himself was considered to be one of the greatest minds AND educators of our time. You really need to see this video and there is a very good chance it will shatter your CURRENT view of what 'nondeterministic' means (not to say that a more mature view of nondeterminism doesnt exist.. one indeed does)

An example of how deterministic the double-slit results are can be found on many credit cards, childrens trading cards, many national currencies, as well as stickers --> The old favorite of marketing --> Holograms!

1979!
Ok, he's a real smart guy, and a gifted speaker, but 1979!
The non-locality experiments were done in the 90's and have given great credence to the notion that we simply don't understand everything (yet). This is because, with our current ways of thinking, we cannot explain how the impossible is demonstrated to be true.
Holograms are the interaction of 2 or more photons (or whatevers) while the double-slit experiment is the interaction of 1 photon with... hey! What's that photon interacting with? Stop that you bad photon! (we haven't figured it out yet, but if Quantium Electrodynamics has an explaination, I'm all for it! BUT most of my research was done after 1979, and the double slit was never considered "solved" that I've ever seen.)

Edit: I get what you're saying atreas, and it makes sense to me But i doubt certain other posters will 'get it'! Lol!
Must look up those 3 experiments later...

Edit Edit: Okay, I watched the first one (1.17 min) and three things are apparent: Newton was really smart! 2) light is a particle. 3) randomness exists. Sorry, no causality because causality cannot include the random.
Oh, at least in this first lecture, he doesn't talk about the double slit at all...

 

[NapoléonPremier]

@Brighteye
THAT is why I get so frustrated. Rather than addressing and discussing an issue, you slip and slide around things, all the while suggesting that your opponents are the ones being illogical. Your refusal to address an issue is the worst form of "logic' and the lowest form of "discussion" imaginable.

Wow, the cat has turned into a tiger...

 

[NapoléonPremier]

One could give all the believers a headache by asking them when exactly a "soul" gains control of a human - Immediately after the sperm breaks into the egg? Some time while its a fetus? The moment the baby leaves its mothers body? When the cord gets cut? When its spoken its first word? When its taken its first step? Its first kiss? When it procreates? Just before death?

FYI - but only as entertainment - the person (ie the "soul) enters the body at the very first moment it starts existing - meaning your first suggestion, "immediately after the sperm breaks into the egg". However, during gestation the fetus is floating in the amniotic liquid, somewhat suspended and still constructing itself, so during that time the "soul" kind of floats around the body. It's only at the end, when the actual birth sets off and the baby breaks his "cocoon" and starts heading inside the mother's belly towards the exist that the soul really gets tied down to the body and makes the experience of the physical - and this is also one of the most traumatic moments of the human existence, a trauma whose consequences will actually extend to the entire life of the person - but that's another story. And actually, the very first trauma, the "mother of all trouble" is that very first moment you mentioned when the "soul" enters the newly formed egg and realises that, like oil and water, matter and spirit don't really mix. It's gonna create the first thoughts of the persons (negative ones), which will turn into very strong negative emotions with the second step (the "passage" at the moment of actual birth). This creates an extremely negative mental process, universal, that every person carries in them from the beginning on, and whose effects are generally known as the "human condition" (sadness, guilt, feeling of being lost etc.).
I know, I know... Don't worry. You will.

 

[NapoléonPremier]

I can accept whatever you want - that soul doesn't exist or that it exists. Still, I have a small scientific problem - I notice that there is a medical (i.e. scientific) specialty that's called "psychiatrist". Of course, psyche = soul (even if you doubt that, I can tell you that from the root of the word it can't be different: psyche is the soul in Greek). That means, there is something that science thinks it exist, and also that is different from "neurology".
I always wondered not so much whether soul exists or not (that can be easily settled on a personal level), but what exactly the science was trying to cure - something that they can't know whether it exists? And if they couldn't know that, how did they created a specific medical field?

Now you're being a little sophistic - but from an honorable Greek man this isn't completely surprising... As I'm sure you know, even if the root of the word is "soul", the modern meaning of the word is "mind", and too many psychiatrists this latter word is understood only as the series of chemical processes that take place in the brain.
And btw, psychiatrists haven't ever cured anybody. They can just somewhat control the symptoms through heavy medication. If you study this field in a little detail, you'd be appalled to discover how little they know about how the human mind works - let alone the "soul" - and thus how powerless they are to cure/change it. For the most part, psychiatry is just a field of classification (this series of symptoms is called "schizophrenia", this one is called "paranoia" etc...). Psychiatry is probably one one the most backwards field of medical science.

 

[NapoléonPremier]

As I was reading the thread (and I was internally translating everything into my language - I can't avoid that) I was always thinking that this whole conversation couldn't have been done in my language, because we don't have two words for the "soul". In my language a human being is called by definition "enpsychon on", which means "a being that has a soul". That of course doesn't prove the existense of soul - just shows the beliefs of people at the ancient times when my language was created.

Do you HAVE to love pubescent boys to be a good philosopher ?

 

[NapoléonPremier]

@jar2574
I don't think you understand the meaning of Occam's Razor one bit.
Occam's Razor is a principle we use to weed out theories which are likely not to be true. It states the following: The least complicated explanation is likely to be true.

Ok : the world does not really exist. It's all just one big illusion. How simpler than that can you get ?

 

[Dusty Monkey]

1979!
Ok, he's a real smart guy, and a gifted speaker, but 1979!
The non-locality experiments were done in the 90's and have given great credence to the notion that we simply don't understand everything (yet).

What do the non-localiy experiments have to do with the double-slit experiments? NOTHING!!!!!!!

DO NOT CHANGE THE SUBJECT

You stated, and I quote you, that the double slit results "defies all understanding"

WRONG! PERIOD! W R O N G !

Changing the subject doesnt make you right.

And on top of it all, attacking the date of the knowledge only makes your stand worse. You apparently don't even know what we knew in the 1950's, let alone now!!!! Your statement on the subject of the double-slit experiment were MADE UP!!!! YOU MADE IT UP!!!! PERIOD!

I tried to let you off easy on that but now its too damn bad... you now have to deal with it.

DO NOT MAKE THINGS UP

Holograms are the interaction of 2 or more photons (or whatevers) while the double-slit experiment is the interaction of 1 photon with... hey!

WRONG. The hologram effect does not require "2 or more photons" - It has nothing to do with the interaction of photons between each other. It has to do with the property of SINGLE photons.. and if you had watched more than the 1st lecture you would have known this!

STOP MAKING THINGS UP!

Edit Edit: Okay, I watched the first one (1.17 min) and three things are apparent: Newton was really smart! 2) light is a particle. 3) randomness exists. Sorry, no causality because causality cannot include the random.
Oh, at least in this first lecture, he doesn't talk about the double slit at all...

If you had watched the 1st lecture, you would know that randomness in this case has nothing to do with causality. Randomness in this case has to do with *our* (as in SCIENCE) ability to predict. Period.

You are apparently making things up now in an effort to save yourself from admitting that causality has not been disproven. Very very sad.

 

[NapoléonPremier]

No scientist would claim that invisible monkeys were irrelevant to gravity, because he would realize that though monkeys are probably are irrelevant to gravity, he could not prove that the monkeys did or did not exist.
Scientists stick to things that can be tested.

It's called the "principle of refutability", and it was formulated by austrian philosopher/epistemologist Karl Popper.

 

[NapoléonPremier]

I'm hoping (cross my fingers) that A Scanner Darkly will prove worthy of the novel.

(/Philip K. Dick)

Yeah, although I have some reservations about the style they chose, ie real movie with real actors (Keanu Reeves etc.) THEN computerized into animation. I would have prefered a real movie. More generally, I think the abuse of the whole "effects" fad these days tends to weaken the story instead of strengthening it. And this is even more true with science fiction. I mean the new Star Wars (Episode 1,2,3) look like a big video game crap compared to their ancestors made 20 years earlier by "tinkering".

 

[NapoléonPremier]

Atreas corrected me about the Greek use of the word 'psuche'. I was startled that my memory was so bad, so I looked it up.
Psuche (in ancient Greek, not modern Greek) does not have to mean soul.
http://www.perseus.tufts.edu/cgi-bin/ptext?doc=Perseus:text:1999.04.0057:entry=#115888

Check out meaning 4. Mind/personality.

Meaning 3: soul

Meaning 1: life!

This is pretty fascinating : for the Ancient Greeks, life/soul/mind were just different manifestations of the same thing. This is something to think about...

 

[NapoléonPremier]

Sure, if I saw any evidence of any sort that souls exist, I'd consider it. I don't see how I can investigate the subject any better than a scientist working in a lab could, though.

Well at least for now, I doubt that any scientist working in a lab would find it. However, should you be interested, the so called "supernatural" ability of telecrux - ie the ability to get out of your body to see things not in your immediate physical surroundings - is actually not that hard to achieve. And then you definitely know that you are more than your body. You don't have to "believe" it.

Apparently quantum physics makes certain statistical predictions that would be violated if hidden variables existed.
I got the following from Wikipedia:

Thanks The detail of the experiments seems pretty complex though, I guess you'd need to really spend some time to get the ins and outs. But it would be time well spent.

And btw, since you seem to be on the "very rational" side of the fence, doesn't it rattle you a little that the very basis of everything (the quantum level) would be something undeterministic ?

 

[NapoléonPremier]

Your conclusion was not necessarily true, however. Not all of the premises have been tested. No one has tested whether all intelligence is created solely by physical things that are governed by causality. It's untestable.
This is just another demonstration of why science has its limits. We can't test everything.

@ Brighteye :

To elaborate on your Gedankexperiment (which has already caused more trouble than actually replicating a brain for real ) :

do you think that if we could get a "virgin brain", and replicate in it all the neural cablings from the brain of another person (in exact detail), we would get the exact same person, with the same personnality, same memories, same beliefs, etc. ?

I think it might be possible to transfer all these things. But would the resulting 'person' be the 'same'? I'd say no. Even if all the data leading up to a point in time might be the same, the choices made after that point could well be different.
Just like the double slit experiment (ie: it defies our ability to predict the outcome)

You've missed a few important factors out, such as ion and transmitter/hormone distribution, quantity of cellular stores of these, density of receptors....

But yes, if we did create an identical brain, it would have the same personality, memories etc. It would have a different soul.

I think that my proposed experiment - which is not a "thoughtexperiment" - could actually allow to test two key concepts that came about repeatedly along this thread : "souls" and freewill. If an exact physical replica of an existing brain would give a "being" with the exact same personality and charateristics than the owner of the original brain, then I think it would prove pretty clearly that souls don't exist and that all human characteristics are the result of chemical processes and can be duplicated by duplicating these same chemical processes.
And, if you put the "original subject" and the replicated brain in the same environment (but separate) and subjected them to different choices, you could see if they would make the same choices every time, thus giving a pretty good indication about the existence of freewill or not (of course it wouldn't be absolute proof, since all parameters wouldn't exactly match - the original person has a body, the two environments could only be close approximations etc...)
Of course, we're very far from being able to doing this today (as far as I know), and maybe we won't ever, but I think this would be an extremely interesting experiment which would yield some hard data on these subjects (souls, freewill), and many others. Plus it would be massively fun...

 

[kingjoshi]

And btw, since you seem to be on the "very rational" side of the fence, doesn't it rattle you a little that the very basis of everything (the quantum level) would be something undeterministic ?

Ignoring the fact that the question wasn't aimed at me and I've skipped most of the thread...

Why should it matter? Whether the foundations of everything are deterministic or not, it doesn't diminish our capabilities in making predictions based upon physical laws that have been derived over human history. Our knowledge and understanding can only grow, so what's the problem?

As we learn more about ourselves and the world, the AIs we create will be better. We'll eventually have AI that's indistinguishable from other humans on the Internet. We'll create robots with AI that will learn something no human has learned or thought before.

The timeframe is unknown. Who knew when Turing set the stage that we'd only progress this far in the grander scheme? As much as we've learned since then, sometimes our task seems even more daunting now. Yet progress has been made and we'll continue to inch closer.

 

[kingjoshi]

I think that my proposed experiment - which is not a "thoughtexperiment" - could actually allow to test two key concepts that came about repeatedly along this thread : "souls" and freewill. If an exact physical replica of an existing brain would give a "being" with the exact same personality and charateristics than the owner of the original brain, then I think it would prove pretty clearly that souls don't exist and that all human characteristics are the result of chemical processes and can be duplicated by duplicating these same chemical processes.
And, if you put the "original subject" and the replicated brain in the same environment (but separate) and subjected them to different choices, you could see if they would make the same choices every time, thus giving a pretty good indication about the existence of freewill or not (of course it wouldn't be absolute proof, since all parameters wouldn't exactly match - the original person has a body, the two environments could only be close approximations etc...)
Of course, we're very far from being able to doing this today (as far as I know), and maybe we won't ever, but I think this would be an extremely interesting experiment which would yield some hard data on these subjects (souls, freewill), and many others. Plus it would be massively fun...

Freewill is ill-defined. What would your objective definition of freewill be for the experiment? How would it translate over to comparisons of people of different DNA,experiences, etc?

Even from an objective viewpoint, I don't see how a soul is disproved from your argument. Identical twins with the same DNA hasn't made a dent in the notion of souls. Even in vitro fertilization, cloning of embyonic cells, etc: it doesn't matter.

Souls are also ill-defined. Being such, it can be a moving target. Regardless of what tests or experiments you perform, rationalization can always be done regarding the supernatural (heck, people choose to disbelieve physical evidence in front of them. For example, the Mormon church believes that Native Americans are descendents from Jews. And their own tests have shown that there is no connection in the DNA. Yet the church still believes.).

 

[NapoléonPremier]

I understand the meaning of Ockham's Razor.
It can be expressed in Latin as, "entia non sunt multiplicanda praeter necessitatem."
In English this means, "entities should not be multiplied beyond necessity."

Isn't introducing Occam's razor already "multiplying entities beyond necessity" ?