Can Artificial Intelligence Exist? Investigating the Possibilities

[chris_tams]

Can an artificially intelligent machine exist. One that will pass a Turing test?
Does a brain work algorithmically, or is there really something we call a "spirit" inside of us that generates a personality?
Or could a computer potentially reproduce the way in which a human brain works by using computer simulated subatomic particles to represent every single particle in a human brain?
Would this result in artificial intelligence?

[yes youth]

 

[hitssquad]

Can an artificially intelligent machine exist

Can an artificially intelligent machine exist. One that will pass a Turing test?

Theoretically, humans themselves are examples of artificially intelligent creatures. So, by virtue of the fact that humans exist, the case of an artificially intelligent machine existing has already been demonstrated.

Does a brain work algorithmically, or is there really something we call a "spirit" inside of us that generates a personality?

The spirit would have to exist outside of the universe, not inside of us or anywhere else within our universe. There would need to be a spirit-channeling place within us, however.

The answers of whether there are or are not spirits channeled by us are untenable either way, by definition of spirits existings solely outside of our universe. There may be indirect ways of detecting any given influence they may be having, however. Given that they would not be subject to our laws and could therefore do anything without any cause coming from our universe and furthermore that there is no way to establish cause and effect regarding their actions, their actions within our universe would have to be exactly purely random. There would have to be an exact zero correlation between channeled spirit influence in our world and anything else that can be measured in our world.

This might be establishable either way (that in our universe there is such a thing as pure random things or that there is not) via statistical instruments, and if it is then we would have either established or not established the existence of spiritual influence in our world - up to the point of the statistical power of the given statistical instruments employed.

Or could a computer potentially reproduce the way in which a human brain works by using computer simulated subatomic particles to represent every single particle in a human brain?

This would not be necessary any more than neutron transport computer simulation needs to simulate the look and feel of real neutrons. The results of thinking are reliably statistical and it is these statistical results that need to be simulated. The underlying processes only need to be simulated as far as they form part of the statistical results we are looking for.

Would this result in artificial intelligence?

No. Simlulating the end effect of thinking is both sufficient and the only way to effectively simulate human thinking (given that human thinking is what is desired under the label of artifial intelligence).

Artificial intelligence pioneer John McCarthy has some more thoughts and Q&A's on artificial intelligence.

 

[chris_tams]

Surely the most accurate way of simulating statistical results would be to simulate the underlying processes. If we were to purely analyse the way in which these results are produced, and then simulate that through some algorithmic process, this would not take into consideration other "animal" tendancies, which result in a life-form being classed as "intelligent".

 

[Imparcticle]

Theoretically, humans themselves are examples of artificially intelligent creatures. So, by virtue of the fact that humans exist, the case of an artificially intelligent machine existing has already been demonstrated.

By definition, "artificial" indicates something that is "Made by humans; produced rather than natural". Humans are created naturally not by some other unnatural way. Would you consider "test-tube babies" as artificial or natural?

The spirit would have to exist outside of the universe, not inside of us or anywhere else within our universe. There would need to be a spirit-channeling place within us, however.

How do you substantiate this claim?

The answers of whether there are or are not spirits channeled by us are untenable either way, by definition of spirits existings solely outside of our universe.

What definition are you going by? According to "dictionary.com", a spirit is defined as thus:
"spir·it n.

The vital principle or animating force within living beings.
Incorporeal consciousness.
The soul, considered as departing from the body of a person at death "

There may be indirect ways of detecting any given influence they may be having, however. Given that they would not be subject to our laws and could therefore do anything without any cause coming from our universe and furthermore that there is no way to establish cause and effect regarding their actions, their actions within our universe would have to be exactly purely random. There would have to be an exact zero correlation between channeled spirit influence in our world and anything else that can be measured in our world.

Where do you get this information?

 

[loseyourname]

A spirit, by definition, has no spatial extent. I think that it what he means when he says it would exist outside of the universe. I see no reason why it shouldn't have any temporal extent, however, so his point may be moot.

 

[p-brane]

Everything we think is artifical is a product of nature. Therefore, there is nothing that can be called artifical. My two cents.

 

[Mentat]

Can an artificially intelligent machine exist. One that will pass a Turing test?
Does a brain work algorithmically, or is there really something we call a "spirit" inside of us that generates a personality?
Or could a computer potentially reproduce the way in which a human brain works by using computer simulated subatomic particles to represent every single particle in a human brain?
Would this result in artificial intelligence?

[yes youth]

Chris, your first question lumps together the concepts of an intelligent machine, a sentient machine, and a machine that can pass the Turing test. Granted, these are closely related concepts, but they are certainly not equivalent.

You see, the Turing test is designed to be a positive test of sentience/consciousness. Only sentient beings can pass the test, but failure of the test says nothing of the subject's consciousness or intelligence.

Yes, a machine can be built that will be intelligent. A PC is, by most definitions, intelligent. Also, as has been pointed out, the human being is an intelligent computer, and thus the concept is clearly possible in principle at the very least.

Yes, a machine can be designed that can pass the Turing test, and thus prove its sentience. However, a machine could be built that was sentient but could not pass the test. Also, there is reason to believe that a machine cannot be designed to, from the beginning of its existence, immediately pass the Turing test. Dennett and others have postulated that a being cannot be sentient until it has experiences over a period of time. It needs to interact with the world; needs to "feel it out", so to speak. What is necessary for passing the Turing test is "world knowledge", and this can only be acquired through having interacted with the world, and getting an intrinsic "feel" for the way it works. There is a good essay on this, by Dennett, in the book "Brainchildren: essays on designing minds". I highly recommend it.

Now, to your second question. I, personally, like the selectionist/Dennetian concept of an algorithmic construct, based on much "stupider" homunculi, but that is my opinion based on what I have read. There could exist a "spirit" of sorts, but then one must ask oneself if that "spirit" is physical. If it is, then it would be physically detectable (and you'd have to ask yourself how it gets its personality in the first place, unless it itself has a spirit...which would lead to infinite regress...or you could just say that it has a personality of its own, with no need of a higher entity -- in which case I would wonder why we assume its existence in the first place, since we could just as easily have a personality of our own, just as we say it does), and we'd have to explain why we haven't detected it. If it is not physical, then we have to explain how it interacts with the physical brain (since clearly this spirit would have no influence or importance if it did not interact with the brain somehow). You'd need some bridge between the two (the spiritual body, and the physical body), and the bridge would have to be both non-physical (to interact with the spiritual) and physical (to interact with the physical), which is illogical.

So, no, I don't think there is a spiritual side to human consciousness/personality, because I don't see how it would work, or what necessary purpose it would serve.

As to your third question, hitssquad is probably right. It is not necessary to simulate all the particles that compose a brain, in order to produce a human form of consciousness. After all, the resultant algorithm of functions is really all you're after in producing human consciousness. A completely different algorithm of completely different functions could produce the same effect (in terms of sentience of the being), and so there is probably a very wide array of different materials and different functions that can become sentient.

 

[Mentat]

Everything we think is artifical is a product of nature. Therefore, there is nothing that can be called artifical. My two cents.

But, so long as it is produced by humans, it is definitively "artificial", right? Ergo, all artificial things are natural, but they are also artificial.

 

[p-brane]

But, so long as it is produced by humans, it is definitively "artificial", right? Ergo, all artificial things are natural, but they are also artificial.

Thats the confusing way of looking at it, Mentat. That is, relying on a definition of a state made by the creators of the state that pertains to the nature of the state they made.

In other words, the baker calls 11 buns a "baker's dozen" and you think you're buying 13 buns. Do you believe the baker or do you believe the amount of buns in the bag?

When I see a Zyrcon diamond or an artifical heart or artificial lighting, I see that nature has come a very long way with its complexity which affords its precise and perfect delivery of simplicity. That's my opinion.

 

[Mentat]

Thats the confusing way of looking at it, Mentat. That is, relying on a definition of a state made by the creators of the state that pertains to the nature of the state they made.

In other words, the baker calls 11 buns a "baker's dozen" and you think you're buying 13 buns. Do you believe the baker or do you believe the amount of buns in the bag?

Not a fitting analogy, since the meanings of "artificial" and "natural" are not mutually exclusive (which was my point to begin with), but the numbers 11 and 13 are.

Artificial is but a sub-set of natural, since all artificial things are natural (having been produced by natural beings (humans) through natural processes (the only ones available to humans in the first place)) but not all natural things are artificial. It's like saying "no, that's not an animal, it's a beetle". Perhaps beetles aren't what we usually think of when we think of "animals", but they certainly are animals.

 

[gravenewworld]

I don't know anything about computer science/programming but I do know that Godel's proof is used to often argue that humans will always be superior to a computer.

 

[Mentat]

I don't know anything about computer science/programming but I do know that Godel's proof is used to often argue that humans will always be superior to a computer.

What do you mean by "superior"? What do you mean by "computer"? Also, could you expound on the point about Godel? Do you mean that there's something in the Incompleteness Theorems that is somehow related to consciousness?

 

[selfAdjoint]

This is a version of Penrose's argument in his The Emperor's New Mind. Goedel showed that systems capable of proving arithmetic necessarily contain propositions that can both be proved true and untrue. Such propositions are called undecidable. Turing and others showed that digital computers are in the class of systems that have undecidable propositions.

Now Penrose told us that some of his friends had discovered a particular undecidable proposition in arithmetic, and claimed that he and they could see that it was in fact not undecidable but true. This he ascribed to human minds being non-goedelian and able to find truths which machines, being goedilian, cannot. Hence humans are forever superior to machines in this respect.

But Penrose's thesis met a storm of criticism from mathematicians who have spent their lives in studying Goedel's theorem and its consequences. And the notion that Penrose and his pals could intuit the truth of a theorem - a very abstract theorem - was regarded as just self love. If they could do that, why have proofs in the first place? In his subsequent books Penrose has quietly abandoned his Goedel theory.

 

[Ethanol]

Did You all see the new picture I,Robot. A creator will never creat something superior to himself. Besides u must all have the idea of how a human brain works in reflexes. How does the brain work when you are doing homework? How does logic work? Simple. Their is something superior in humans that machines haven't. This thing could be called The Vital Force (not like the vital force of organic compounds). If anybody can reply with a better theory please help me.

 

[chronon]

This is a version of Penrose's argument in his The Emperor's New Mind. Goedel showed that systems capable of proving arithmetic necessarily contain propositions that can both be proved true and untrue.

I hope not - that would imply that arithmetic is inconsistent (and that anything can be proved). Gödel showed that there are propositions which can't be proved either way using the axioms of arithmetic. Essentially he encoded the statement "this statement is unprovable" and so came up with a true but unprovable statement. There is the related Halting Theorem for computers, which says that no general program can be written which takes computer programs as input and decides whether they will halt.

Roger Penrose, and earlier John Lucas used this to argue that there is something special about how humans recognise truth, which is more than a machine could ever do. I find these arguments doubtful.

One trouble is that although people like discussing this endlessly, it is very hard to find any serious work done on trying to finds the limits of the Halting Theorem. What if you tried to design a Halting program which knew about the halting theorem and could recognise you were trying to trick it? Is it doomed to fail somehow, or do you find that the machine has just the same limitations as humans as in "Penrose cannot consistently assert that this statement is true"?

 

[Hurkyl]

I think selfAdjoint was referring to another theorem of Godël's:

No sufficiently powerful consistent theory (i.e. one containing number theory) can prove itself consistent.

Since it cannot prove itself consistent, it cannot prove number theory consistent. (Though it can prove that number theory is consistent relative to itself)

 

[selfAdjoint]

I think selfAdjoint was referring to another theorem of Godël's:

No sufficiently powerful consistent theory (i.e. one containing number theory) can prove itself consistent.

Since it cannot prove itself consistent, it cannot prove number theory consistent. (Though it can prove that number theory is consistent relative to itself)

The goedelian proposition that Penrose used in The Emperor's New Mind was actually a metamathematical theorem involving infinite sets of numbers.