Can Artificial Intelligence Ever Match Humans?

[npts2020]

Sione; Emotions are predictable. Advertisers, politicians, preachers and interrogators (among many) know this and use it to their advantage. The fact that nobody can predict every single emotional state in a subject does not mean that emotion is not deterministic, only that there wasn't enough information to succeed in doing so. Exactly the same thing can be said about trying to predict the actions of a completely rational subject. Even if both are completely deterministic, it doesn't mean that any human (and by extension computer) will ever have enough information to predict every outcome. The problem is in knowing every variable in a highly dynamic system.

[Sione]

bascule,

 

That's great. I agree with almost everything you said.

 

 

There is no evidence of quantum behavior in the brain. Moreover' date=' there has been substantial research into the behavior of the brain and models constructed which rely on classical mechanics. Neurophysiologists have not discovered any brain behaviors which cannot be explained using classical mechanics.

[/quote']

 

I especially agree with this.

 

Artificial Neural Networks might be a good candidate to simulate "true emotion/thought/consciousness". But to answer the question fully and be sure if it is really "real", we must know the mechanics of thought process or otherwise be able to compare it with our AI. Do you know how random and spontaneous are Artificial Neural Networks?

 

 

I insist -determinism- is fundamental property to implementation of algorithm, or physical structure, for AI.

 

 

As an emergent materialist' date=' I would argue that qualia, emotion, etc. would emerge from a complete enough functional simulation of the human brain.

[/quote']

 

I agree, but how deterministic is something which 'emerges spontaneously'?

 

 

Dennett argues the underlying mechanics of this process are still deterministic, i.e. given the same mental states and being put in the same situation we will make the same choice every time.

 

That's fine, I do not insist. All I'm saying is that IF it turns out that it is not deterministic, say because of quantum mechanics, then how do you simulate it short of building a physical replica?

 

My only point is:

- WITHOUT the DEFINITION of underlying MECHANICS there can be no TRUE SIMULATION.

---

Merged post follows:

Consecutive posts merged

Sione; Emotions are predictable. Advertisers' date=' politicians, preachers and interrogators (among many) know this and use it to their advantage.

[/quote']

 

Sorry, we are talking about simulation, algorithms and determinism here, you are talking about statistics and probability.

 

 

The problem is in knowing every variable in a highly dynamic system.

 

Yes' date=' but first you need to ask if it is at all POSSIBLE to know. If it is based on Quantum Mechanics, then it is not possible to know, it is NOT DETERMINISTIC in that case. The first thing to know about simulating something is MECHANICS of it, do you know mechanics of thoughts and emotions? If not, then you can not claim they are either deterministic or not, only statistically, which is fake. [b']Forget the word "predictable", [/b]it is wrong word that apparently makes confusion. This is what I am talking about:

 

de⋅ter⋅min⋅ism –noun

1. Describes a system whose time evolution can be predicted exactly. In contrast to probabilistic.

2. Describes an algorithm in which the correct next step depends only on the current state.

 

 

 

Let me suggest something then.

 

You' date=' kind of, did not answer to my questions, but keep challenging me for some strange reason. I hope what you're going to say next has anything to do with anything, lets see...

 

 

If you measure the spin of an electron in an unknown state, you can measure the spin as up or down. Call the up 1 and the down 0. This is considered to be a non-deterministic result.

 

We do not know something before we know it, ok.

 

 

Now' date=' flip a coin and call heads 1 and tails 0. This is known to be a deterministic result.

[/quote']

 

I see what you mean, but you are confusing yourself.

 

It is not about result and reading, but about underlying physics. Classical physics is deterministic, quantum physics is not.

 

 

 

Finally' date=' I shall call the flip of the coin a simulation of the measurement of the spin of the electron. Now, can you tell the difference, from looking at the 1s and 0s, which is the simulation and which is not?

[/quote']

 

No, I can not. And, your point is?

 

However, if I am programmer to simulate AI for you, then you tell me, will you accept it to have "true emotions" if I fool you the same way? My point is about defining TRUE emotions, not about statistical fakes and Pokemon.

 

- WITHOUT the DEFINITION of underlying MECHANICS there can be no TRUE SIMULATION.

 

 

If you cannot' date=' then I just gave you an example of a deterministic simulation of a non-deterministic thing. How many individual quotes do you think you will have to break this paragraph into for you to think you have given any sort of counterargument?

[/quote']

 

Why would I want to counterargument that, it has nothing to do with anything I said. People can be tricked by statistical simulation, that's what I said. The question is if you accept that as "true" and "real emotion". This is not about being tricked, but about true nature and possibility of real simulation, stop fooling yourself and pay attention!

Edited January 28, 2009 by Sione
Consecutive posts merged.

[bascule]

Artificial Neural Networks might be a good candidate to simulate "true emotion/thought/consciousness". But to answer the question fully and be sure if it is really "real", we must know the mechanics of thought process or otherwise be able to compare it with our AI. Do you know how random and spontaneous are Artificial Neural Networks?

 

Artificial neural networks have effectively plateaued. The only focus of present research is narrow AI applications.

 

I insist -determinism- is fundamental property to implementation of algorithm, or physical structure, for AI.

 

Anything which runs on a Turing machine is inherently deterministic.

 

I agree, but how deterministic is something which 'emerges spontaneously'?

 

The emergence is of a secondary metaphysical construct, in this case your perceived conscious experience. If you believe emergent materialism, the content of this experience is entirely determined by the physical goings on of the brain.

 

That's fine, I do not insist. All I'm saying is that IF it turns out that it is not deterministic, say because of quantum mechanics, then how do you simulate it short of building a physical replica?

 

If there are quantum effects at play somewhere their probabilities would need to be modeled and simulated.

 

I'm sure it's not too terribly difficult for quantum physicists to build models of quantum effects which have similar statistical properties to the real thing.

 

My only point is:

- WITHOUT the DEFINITION of underlying MECHANICS there can be no TRUE SIMULATION.

 

In the case of the brain, the underlying mechanics are classical (chemistry, classical field theory, etc) and are extremely well defined and understood

 

New MRI and other neural imaging technologies are making it increasingly easier to build sophisticated maps of the structure of the brain. These can in turn be applied towards creating a full computer simulation.

 

Alternatively we could model the entire human development process inside a computer, starting with a fertilized egg and letting it mature into a complete artificial human.

[Sione]

I insist -determinism- is fundamental property to implementation of algorithm' date=' or physical structure, for AI.

[/quote']

Anything which runs on a Turing machine is inherently deterministic.

 

I believe we achieved some understanding here.

 

And so, before we start talking about simulating this "machine", we should be aware if it is really deterministic or not.

 

 

------------------

Now, I see you argue brain is "Turing machine" and that is exactly the point we need to discuss first. That is the whole point of my talk about determinism, to discuss if brain is deterministic type of Turing machine or is some kind of non-deterministic Quantum machine.

 

Thank goddess Chance, discussion will be more obvious and straight forward now... so, let us argue. I say, you are wrong!

 

 

The emergence is of a secondary metaphysical construct' date=' in this case your perceived conscious experience. If you believe emergent materialism, the content of this experience is entirely determined by the physical goings on of the brain.

[/quote']

 

I will say what I believe for myself, you tell us what you believe. Are you saying there is no free will, consciousness is illusion, we can not really change anything with will, intention or planning is a mirage, and we are just some puppets of our brain's chemical structure?

 

 

If there are quantum effects at play somewhere their probabilities would need to be modeled and simulated.

 

That is right' date=' they must be modeled if we are to call it a 'real-deal'.

 

However, without knowing MECHANICS, we can not know if there are quantum effects at play. Do you suggest you have a reason to be believe there are no quantum effects at play in thought process? That is quite a lot to say about something human race knows nothing about, especially since brain is made out of these quantum particles.

 

 

In the case of the brain, the underlying mechanics are classical (chemistry, classical field theory, etc) and are extremely well defined and understood

 

Please, you are suggesting memory and thought process exist inside the brain as molecular structures, geometrical imprints? It is well known brain functions with electric impulses and it is almost completely unknown about magnetic fields influences to whole thing... my point is, thought process is not chemical, it would be too slow, but ELECTROMAGNETIC. Do you really mean to dispute this?

 

 

By the way, electromagnetic interaction is not well defined at all, it is not even chaotic, it is truly non-deterministic and uncertain, that is why we have Quantum Mechanics explaining it with voodoo. Do you mean to refute this?

 

 

New MRI and other neural imaging technologies are making it increasingly easier to build sophisticated maps of the structure of the brain. These can in turn be applied towards creating a full computer simulation.

 

Alternatively we could model the entire human development process inside a computer, starting with a fertilized egg and letting it mature into a complete artificial human.

 

That sounds great and I would like to hear more about it.

 

However, we CAN MOST CERTAINLY NOT 'model the entire human development process inside a computer, starting with a fertilized egg', are you crazy? We can not even simulate the simple bonding of two hydrogen atoms, only statistically via Quantum Mechanics, and that is fake. Any more than couple of atoms in such simulation and QM will struggle, because it does not have equations for interaction, but equations of statistical geometry description.

 

 

Chemistry is an approximation, think outside the box and look at the big picture... accept this knowledge, do you accept?

Edited January 28, 2009 by Sione

[iNow]

Sione - All neural activity is chemoelectric. If you wish to argue otherwise, I'd like to see citations in support of your position that ion exchange has no impact on signal transduction, action potentials, and neural cascades. Until you share citations in support of your statement:

 

the point is thought process is not chemical, it would be too slow, but ELECTROMAGNETIC.

 

...then you are simply making things up and your words can be disregarded.

 

 

 

Basically, you introduce a false dichotomy, as it's the chemical reaction which causes the electromagnetism on which your point rests.

 

 

Oh, and what does this have to do with computer science?

[Sione]

Oh' date=' and what does this have to do with computer science?[/quote']

 

What are you referring to, what confuses you?

 

Simulating thought process has to do with computer science, what part do you not understand?

 

 

Sione - All neural activity is chemoelectric. If you wish to argue otherwise' date=' I'd like to see citations in support of [b']your position that ion exchange has no impact on signal transduction, action potentials, and neural cascades.[/b] Until you share citations in support of your statement:

 

the point is thought process is not chemical, it would be too slow, but ELECTROMAGNETIC.

 

...then you are simply making things up and your words can be disregarded.

 

chemoelectric? Hahahaa...

 

Humans are so persistently oblivious to magnetic forces, oh mercy!

This confirms my point from another thread about Lorentz... remember, there is no electricity without magnetism.

 

 

Chemistry is an approximation of electromagnetics, can you grasp? Your statement that something is chemoelectric, but not electromagnetic is an insult to human race, you should be put in jail for saying such nonsense. "ion exchange" IS ELECTROMAGNETIC INTERACTION, thought some might learn about it at chemistry class, sorry if that confused you.

 

 

If you are not aware of electromagnetic properties of human brain, then you're not even worth making fun of. I pity you, and looking forward to see what others will say about that. In the meantime, you could do some reading and save yourself further embarrassment.

Edited January 28, 2009 by Sione

[iNow]

What are you referring to, what confuses you?

 

 

 

chemoelectric? Hahaha.

 

Chemistry is an approximation of electromagnetics, can you grasp? Your statement that something is chemoelectric, but not electromagnetic is an insult to human race, you should be put in jail for saying such nonsense. "ion exchange" IS ELECTROMAGNETIC INTERACTION, thought some might learn about it at chemistry class, sorry if that confused you.

 

 

If you are not aware of electromagnetic properties of human brain, then you're not even worth making fun of. I pity you, and looking forward to see what others will say about that. In the meantime, you could do some reading and save yourself further embarrassment.

 

Yeah, I didn't figure you were adult enough to share a citation in support of your assertion. Here comes that "disregarding your posts" response about which I previously spoke. Enjoy.

 

 

For our other readers who wish to correct their own understandings and erase Sione's falsehoods from their memory, you can start at links like this:

http://en.wikipedia.org/wiki/Chemical_synapse

 

 

Chemical synapses are specialized junctions through which neurons signal to each other and to non-neuronal cells such as those in muscles or glands. Chemical synapses allow neurons to form interconnected circuits within the central nervous system.

They are thus crucial to the biological computations that underlie perception and thought.

 

<...>

 

Here is a summary of the sequence of events that take place in synaptic transmission from a presynaptic neuron to a postsynaptic cell. Each step is explained in more detail below. Note that with the exception of the final step, the entire process may run only a few tenths of a millisecond, in the fastest synapses.

 

1. The process begins with a wave of electrochemical excitation called an action potential traveling along the membrane of the presynaptic cell, until it reaches the synapse.
   
   
2. The electrical depolarization of the membrane at the synapse causes channels to open that are permeable to calcium ions.
   
   
3. Calcium ions flow through the presynaptic membrane, rapidly increasing the calcium concentration in the interior.
   
   
4. The high calcium concentration activates a set of calcium-sensitive proteins attached to vesicles that contain a neurotransmitter chemical.
   
   
5. These proteins change shape, causing the membranes of some "docked" vesicles to fuse with the membrane of the presynaptic cell, thereby opening the vesicles and dumping their neurotransmitter contents into the synaptic cleft, the narrow space between the membranes of the pre- and post-synaptic cells.
   
   
6. The neurotransmitter diffuses within the cleft. Some of it escapes, but some of it binds to chemical receptor molecules located on the membrane of the postsynaptic cell.
   
   
7. The binding of neurotransmitter causes the receptor molecule to be activated in some way. Several types of activation are possible, as described in more detail below. In any case, this is the key step by which the synaptic process affects the behavior of the postsynaptic cell.
   
   
8. Due to thermal shaking, neurotransmitter molecules eventually break loose from the receptors and drift away.
   
   
9. The neurotransmitter is either reabsorbed by the presynaptic cell, and then repackaged for future release, or else it is broken down metabolically.
   
   

 

<...>

 

An electrical synapse is a mechanical and electrically conductive link between two abutting neurons that is formed at a narrow gap between the pre- and postsynaptic cells known as a gap junction. <...> Electrical synapses are found throughout the nervous system, yet are less common than chemical synapses.

 

 

Again though, none of this has anything to do with computer science, and there is a bridge somewhere missing a troll.

Edited January 28, 2009 by iNow

[mooeypoo]

chemoelectric? Hahahaa...

[ ... ]

you could do some reading and save yourself further embarrassment.

This is a debate forum, not a bash party.

 

You are here to debate, not to lecture and definitely not to belittle others who take the time to respond to your claims.

 

Lose the attitude, lose the ad hominem and personal attacks, this is not the right place for it, and it does nothing to improve the quality of your claims.

[Sione]

Again though, none of this has anything to do with computer science, and there is a bridge somewhere missing a troll.

 

This is what you quoted yourself:

-"Chemical synapses allow neurons to form interconnected circuits within the central nervous system. They are thus crucial to the biological computations that underlie perception and thought."

 

Computational simulation is a field of computer science. What part do you still not understand? Simulating emotions and thought process means simulating neurons mechanics.

 

 

The rest you might understand when you realize electrochemical reaction is ELECTROMAGNETIC reaction. What do you imagine you and me disagree about? We are saying the same thing, only you fail to realize that underlying mechanics includes magnetic fields as well.

 

So, do you think that mechanics of neurons is independent of quantum properties and deterministic? Do you think chemical reactions are deterministic?

[Cap'n Refsmmat]

I pity you, and looking forward to see what others will say about that.

They will say you are being annoying.

 

To reiterate what mooeypoo has stated: The point of rational discussion is to educate those involved and hopefully reach some sort of conclusion. Insulting the others involved does nothing to achieve either and merely serves the purpose of irritating everyone involved.

 

If you would like to discuss this subject with iNow, feel free. If you'd like to continue your current style, well, don't.

[iNow]

It would also help if Sione would stop going back and editing his posts long after others have already responded to them.

[Cap'n Refsmmat]

If that is causing problems, vB now keeps track of all edits, so I can look back at old versions of posts.

[Sione]

So, what is the problem with editing, correcting spelling and stuff?

 

You are here to debate' date=' not to lecture and definitely not to belittle others who take the time to respond to your claims.

[/quote']

 

What do you mean "my claims", it is general knowledge from scientific textbooks.

 

All I say is that brain and neurons are ultimately made of subatomic particles and therefore subject to electromagnetic interaction and quantum uncertainty. I say chemical interaction is just an approximation to electromagnetic interaction, do you think this is 'my claim' or current scientific understanding?

 

1.) Do you think thoughts and feeling are not electromagnetic, but chemical?

2.) Do you think memory and thoughts are independent of electrodynamics and only obey chemical or chemo-electrical laws?

 

 

 

iNow,

 

Sorry for editing... and I'm glad you finally understand that neurons and brain mechanics have everything to do with computer science if we want to simulate it, since computational simulation is a field in computer science. I forgive you for asking such a silly question so many times.

 

 

Could you please clear up your position, what is your point:

 

a.) Do you think chemical reactions are deterministic?

b.) Do you think chemistry of neurons is independent of electromagnetic interaction?

Edited January 28, 2009 by Sione

[Cap'n Refsmmat]

If you'd like to look at it from an electromagnetic viewpoint, you'll end up simulating every single electron and proton inside the brain. That is not very easy.

 

Chemical laws are merely uses of electromagnetic laws to explain large-scale interactions, which are more relevant on the scale of something like the brain. So while you could talk about electromagnetism and how that makes atoms bond in the brain, it'd be a lot easier to just talk about chemistry.

[mooeypoo]

What do you mean "my claims", it is general knowledge from scientific textbooks.

Then where are the citations?

 

 

 

 

 

Regardless, I very much hope that the fact you quoted/related to only one sentence of what I said does not mean you didn't read - and will implement - the rest of my post.

[Sione]

Then where are the citations?

 

Citations?

For what? That all chemical reactions are electromagnetic interaction?

 

That does not need to be cited' date=' it is part of any introduction to the subject of atomic structure and chemical bonding.

 

http://en.wikipedia.org/wiki/Chemical_reaction

[i']- "A chemical reaction is a process ... Classically, chemical reactions encompass changes that strictly involve the motion of electrons in the forming and breaking of chemical bonds, although the general concept of a chemical reaction, in particular the notion of a chemical equation, is applicable to transformations of elementary particles, as well as nuclear reactions."[/i]

 

Basically it says it is about bonding and motion of elementary particles, such as electron and proton,

 

http://en.wikipedia.org/wiki/Electromagnatism

- "Electromagnetism is the physics of the electromagnetic field: a field which exerts a force on particles that possess the property of electric charge, and is in turn affected by the presence and motion of those particles.

 

Basically, it says the same.

Quantum Mechanics might say the same as well. Which theory we use as a basis for our AI?

 

 

I would rather iNow give citation that say electric impulses can exist without magnetic fields and that chemical reactions can explain the speed of cognition or persistence of memory, how about those citations? I do not think you can even call it a chemistry when dealing with electricity. Chemistry is about bonding and molecular geometry, kind of "freeze-frame" of what is going underneath. Electromagnetism is what is actually happening and it is about electrons, ions, any moving charges and all that.

 

 

If you'd like to look at it from an electromagnetic viewpoint' date=' you'll end up simulating every single electron and proton inside the brain. That is not very easy.

 

Chemical laws are merely uses of electromagnetic laws to explain large-scale interactions, which are more relevant on the scale of something like the brain. So while you could talk about electromagnetism and how that makes atoms bond in the brain, it'd be a lot easier to just talk about chemistry.

[/quote']

 

I completely agree. And I do not mind if it happens that we can model AI by only simulating chemistry.

 

But hold on, simulating electrical impulses and neurons with Artificial Neural Network does not have much to do with chemistry at all, it does not even have much to do with electromagnetism, but rather with some abstraction as is 'parallel computing'.

 

 

That is all fine, call it chemistry, electromagnetism or ANN, I want to know if it is deterministic in any case?

 

 

Is chemistry deterministic? Is thought process deterministic?

The point is to TEST it. We can test what is deterministic and rational like chess move, but how to test for spontaneity, if it exists? All I'm saying is that we can not even try to simulate it unless we know basics, such as determinism of the system.

 

 

I have no idea how anyone can imagine some chemical structure and position of atoms is what we feel and think. The essence of consciousness is about moving electrons, not about chemical bonding. I do not think the speed of chemical reaction can explain thought process. All we see is neurons firing, but 'thoughts' may be more than electric impulses. There is always magnetic field around moving charge, which is the part we do not model in our Artificial Neural Networks simulations.

Edited January 28, 2009 by Sione

[bascule]

In the case of the brain' date=' the underlying mechanics are classical (chemistry, classical field theory, etc) and are extremely well defined and understood[/quote']

 

Please, you are suggesting memory and thought process exist inside the brain as molecular structures, geometrical imprints? It is well known brain functions with electric impulses and it is almost completely unknown about magnetic fields influences to whole thing... my point is, thought process is not chemical, it would be too slow, but ELECTROMAGNETIC. Do you really mean to dispute this?

 

No, however it's clear you don't understand the forces you are describing can be explained as a combination of chemistry and classical field theory, both of which were mentioned in the post you're responding to.

 

By the way, electromagnetic interaction is not well defined at all

 

Actually, on the scales of something like the brain, it is. It's defined extremely well. That's why we can build things like MRI machines. You might try reading about Maxwell's Equations

 

it is not even chaotic, it is truly non-deterministic and uncertain, that is why we have Quantum Mechanics explaining it with voodoo. Do you mean to refute this?

 

Yes, on the scale of structures like the brain it's well understood. The brain consists of large, complex structures whose behavior can be explained by classical mechanics.

 

However, we CAN MOST CERTAINLY NOT 'model the entire human development process inside a computer, starting with a fertilized egg', are you crazy? We can not even simulate the simple bonding of two hydrogen atoms

 

Well, the point would be to perform the simulation at a higher level than taking into account individual atomic bonds. For example, if you precompute a table for how different proteins fold, you don't need to model their Gibbs Free Energy. There are many potential optimizations which can speed up a smart model, as opposed to one which is blindly trying to simulate cells atom-by-atom.

[Sione]

No' date=' however it's clear you don't understand the forces you are describing can be explained as a combination of chemistry and classical field theory, both of which were mentioned in the post you're responding to.

[/quote']

Feel free to point out my confusion and teach me, I will accept your knowledge.

 

You make statement as if it is proven, as if someone managed to simulate AI based on chemistry and field theory, did this happen? Where do you find support to insist on something that?

 

 

 

Actually' date=' on the scales of something like the brain, it is. It's defined extremely well. That's why we can build things like MRI machines. You might try reading about Maxwell's Equations

[/quote']

 

I know everything about Maxwell's and all the other equations.

 

However, I do not know about any AI project based on Maxwell equations, hmh?

 

 

Yes' date=' on the scale of structures like the brain it's well understood. The brain consists of large, complex structures whose behavior can be explained by classical mechanics.

[/quote']

 

I would like if that was all there is to it.

 

Since all is deterministic and well understood, what exactly is the problem about simulating consciousness or emotions?

 

 

Well' date=' the point would be to perform the simulation at a higher level than taking into account individual atomic bonds. For example, if you precompute a table for how different proteins...

[/quote']

 

That is all fine, except that is deterministic.

Do you think such AI could have a sense of humor, be able to dream, believe in god or commit suicide?

 

Now, MRI machines do not reveal mechanics of thoughts, nor they can confirm such process is deterministic, or not. I see you are saying it is deterministic and that all is well known about it, but have you got any experiments or other evidence for that?

 

 

Do you think chemistry is deterministic?

 

Did anyone actually manage to apply those well known formula and simulate AI with some traces of consciousness?

Edited January 28, 2009 by Sione

[iNow]

I forgive you for asking such a silly question so many times.

Phew. I sure will sleep better tonight, then.

 

 

 

I would rather iNow give citation that say electric impulses can exist without magnetic fields and that chemical reactions can explain the speed of cognition or persistence of memory, how about those citations?

No need for me to share such citations, as that was never my position. I know full well that electricity and magnetism are inherently coupled, and I'm not quite sure where you think I ever gave the impression otherwise.

 

You were talking about neural cascades in biological systems, and you were corrected about them being chemoelectrical in nature. If you're referring to computer models, then that is different, of course, but you were not (unless you edited your work after receiving a response, then your posts will bear this out).

 

As for the "speed of cognition," what does that even mean? There is a conduction velocity in action potentials and neural cascades which ranges from 1 m/s to over 100 m/s, and that is dependent on a few factors, primarily axonal diameter (the larger the diameter, the higher the velocity) and also mylenation. However, that has very little to do with overall cognition, as cognition is more dependent on dendritic connections and the density of the neural web. In short, the more connections and the denser the network, the better and faster the cognition.

 

None of this, though, has anything to do with computer science. We are talking about neurobiology, which has been my point for several posts now.

 

 

I do not think you can even call it a chemistry when dealing with electricity. Chemistry is about bonding and molecular geometry, kind of "freeze-frame" of what is going underneath. Electromagnetism is what is actually happening and it is about electrons, ions, any moving charges and all that.

It has become quite clear that you do not understand the neurobiological systems you are attempting to model. You probably understand the computer systems better than I do, but you need to own up to the fact that there are some very intelligent people at this site, it's not your high school classroom, and you don't know more than everyone else on every topic. You're an annoying poster, and I'm done with you now unless you make some marked changes in your tone and style.

[npts2020]

Sione; Anyone's opinion of what is or is not deterministic is just that. In my life I have seen nothing to make me think the universe and all of the processes therein are anything other than deterministic. The further humanity progresses, the better we are at predicting outcomes. If there is a field of human endeavor where this is not true, I am unaware of it which in my view supports the notion of determinism. There are many things that are easily explained today (computers, space travel, wireless telecommunications to name a couple), that people of only a century ago had no concept. The same will be true of future humans in any but the bleakest of doomer estimates. Whether we will ever get to the point of being able to predict everything in the universe is a different matter, but possibly not insurmountable. Predicting what future humans will be capable of is impossible without knowing the limits of knowledge (who knows that?). IMO the limit of our knowledge limits the ability to see the determinism exhibited in the world around us. I look forward to seeing your explanation for why you think it is impossible that everything in the universe is deterministic.

 

BTW A computer algorithm that took into account every variable in the universe would be the universe (or at least indistinguishable from it).

[Sisyphus]

Since all is deterministic and well understood, what exactly is the problem about simulating consciousness or emotions?

 

Computing capacity, for the time being. I don't know whether there are other reasons.

 

Do you think such AI could have a sense of humor, be able to dream, believe in god or commit suicide?

 

Do you have some reason to believe otherwise (other than an appeal to ridicule)?

[bascule]

You make statement as if it is proven, as if someone managed to simulate AI based on chemistry and field theory, did this happen? Where do you find support to insist on something that?

 

Support for this comes from both neurophysiologists and physicists. For example, this paper from Max Tegmark in response to Roger Penrose's hypothesis that quantum mechanics is involved in brain processes:

 

http://arxiv.org/abs/quant-ph/9907009

 

Based on a calculation of neural decoherence rates, we argue that that the degrees of freedom of the human brain that relate to cognitive processes should be thought of as a classical rather than quantum system, i.e., that there is nothing fundamentally wrong with the current classical approach to neural network simulations.

 

The great mysteries of the brain all involve its higher level structure and the interactions between various brain systems, not the behavior of individual neurons. That much is well understood.

 

Since all is deterministic and well understood, what exactly is the problem about simulating consciousness or emotions?

 

As Sisyphus said, computing power is the main limitation. The BlueBrain project is attempting to simulate the neocortex of a rat. The neocortex is suspected to be crucial to intelligence and consciousness.

 

However, even with one of the world's most powerful supercomputers they are only able to simulate one neocortical "column" of a rat. A rat's neocortical column has 10,000 neurons compared to human neocortical columns, which have 60,000 neurons. A human has nearly a million neocortical columns.

 

To simulate just the neocortex of a human, we'd need a computer that's approximately six million times faster than BlueBrain.

[Sione]

None of this' date=' though, has anything to do with computer science. We are talking about neurobiology, which has been my point for several posts now.

[/quote']

 

Can you finally explain how understanding neurobiology in order to SIMULATE brain functionality has nothing to do with computer science?

 

 

No need for me to share such citations, as that was never my position. I know full well that electricity and magnetism are inherently coupled, and I'm not quite sure where you think I ever gave the impression otherwise.

 

Sure, let me remind you of your silly objection:

the point is thought process is not chemical' date=' it would be too slow, but ELECTROMAGNETIC.

[/quote']

...then you are simply making things up and your words can be disregarded.

 

 

...anyway, I'm glad we agree now.

 

 

You were talking about neural cascades in biological systems, and you were corrected about them being chemoelectrical in nature.

 

No, I was talking about computer algorithm and determinism of the system to be simulated. You were talking about chemistry without realizing what you admitted above, that "electricity and magnetism are inherently coupled".

 

 

...anyway, I'm glad we agree now.

 

 

As for the "speed of cognition," what does that even mean? There is a conduction velocity in action potentials and neural cascades which ranges from 1 m/s to over 100 m/s, and that is dependent on a few factors, primarily axonal diameter (the larger the diameter, the higher the velocity) and also mylenation. However, that has very little to do with overall cognition, as cognition is more dependent on dendritic connections and the density of the neural web. In short, the more connections and the denser the network, the better and faster the cognition.

 

Speed of cognition is "speed of thinking", not necessarily the speed of electric impulses, but the speed with which you can recall memory, speed of animation of dreams or 'rate of change of thought'. In analogy to computer, the speed of cognition is the frequency of information output and the speed of impulses along nerve fibers corresponds to speed of BUS or threading pipeline between processing units.

 

Fell free to point out reference to correct terminology and definitions.

 

 

It has become quite clear that you do not understand the neurobiological systems you are attempting to model. You probably understand the computer systems better than I do, but you need to own up to the fact that there are some very intelligent people at this site, it's not your high school classroom, and you don't know more than everyone else on every topic. You're an annoying poster, and I'm done with you now unless you make some marked changes in your tone and style.

 

It has become quite clear that you do not understand the neurobiological systems we are attempting to model. I most certainly understand the computer systems better than you do, but you need to actually articulate your confusion if you want to learn, or even participate in discussion without embarrassing yourself.

 

 

==========================

 

Sione; Anyone's opinion of what is or is not deterministic is just that. In my life I have seen nothing to make me think the universe and all of the processes therein are anything other than deterministic.

 

That's fine' date=' opinions are fine.

 

However, if we want to SIMULATE some system, then we need more than opinions. WE NEED TO KNOW MECHANICS, that is all I'm saying.

 

 

I can not write simulation of airplane based on opinion that lift is produced by Bernoulli's principle, if that is not real-world fact. I need to know real mechanics of flight, and ALL of it, if I am to ever be able to model it and get simulated plane of the ground, right? While opinions are fine we should always try to turn them into observational fact, by finding the experiment to confirm or refute our opinions and theories.

 

In that respect, I point out that brain is made of quantum particles and as such is subject to Quantum Mechanics and inherent UNCERTAINTY PRINCIPLE. Quantum Mechanics is non-deterministic and everything is made out of it. That is my argument, how do you respond to that?

 

I say nothing is deterministic.

 

Do you think chemistry is deterministic?

This should not be too hard to prove or disprove, eh?

 

 

 

==========================

 

Since all is deterministic and well understood' date=' what exactly is the problem about simulating consciousness or emotions?

[/quote']

Computing capacity, for the time being. I don't know whether there are other reasons.

 

Computing capacity should not be a problem to simulate simple organisms. I'm happy with simulated amoebae, bacteria or virus. Or even some simplified version of, say squid nervous system, perhaps fly, rat or cat?

 

I believe Artificial Neural Networks are capable to simulate very large numbers of neurons, even in real-time. I suppose someone will need to do some Googling about it and bring it here, that should be interesting to see.

 

 

I mean scientific discoveries happen so often in the last 10-20 years that it is even possible someone achieved to simulate conscious and emotional AI while I was trying to explain to iNow how "simulation" is a part of "computer science".

 

 

Do you think such AI could have a sense of humor' date=' be able to dream, believe in god or commit suicide?

[/quote']

Do you have some reason to believe otherwise (other than an appeal to ridicule)?

 

I'm serious, why do you find it funny? It's just about the holy grail of AI to simulate exactly those human attributes, and the very question that was posed in OP. Anyway, I asked first, do you have any reason to believe so?

 

 

 

I have every reason to think otherwise. Determinism is one of them. I'm trying to argue non-determinism is the main characteristic of human emotions and key attribute to spontaneous behavior, such as artistic, crazy, humorous, inventive, passionate, creative...

 

...pretty much everything we understand as "human quality" has to do with spontaneity of the reaction, while reason and logic is the least of recognizable humane attributes, because they are easy to fake, they lack originality. WWII, and the rest of the history, and present, teach us that intellect indeed has not much to do with humanity.

 

hu⋅mane –adjective

1. Characterized by kindness, mercy, or compassion.

2. Marked by an emphasis on humanistic values and concerns.

 

 

Do you think chemistry is deterministic?

Are you saying you find it likely some deterministic algorithm has a chance of producing such human-like behavior?

 

 

 

 

 

==========================

 

Support for this comes from both neurophysiologists and physicists. For example' date=' this paper from Max Tegmark in response to Roger Penrose's hypothesis that quantum mechanics is involved in brain processes:

[/quote']

 

Ok, thanks. I will read that and then prove it wrong.

 

Do you think chemistry is deterministic?

This should not be too hard to prove or disprove, eh?

 

 

Since all is deterministic and well understood' date=' what exactly is the problem about simulating consciousness or emotions?

[/quote']

 

As Sisyphus said, computing power is the main limitation. The BlueBrain project is attempting to simulate the neocortex of a rat. The neocortex is suspected to be crucial to intelligence and consciousness.

 

However, even with one of the world's most powerful supercomputers they are only able to simulate one neocortical "column" of a rat. A rat's neocortical column has 10,000 neurons compared to human neocortical columns, which have 60,000 neurons. A human has nearly a million neocortical columns.

 

To simulate just the neocortex of a human, we'd need a computer that's approximately six million times faster than BlueBrain.

 

I thought, just before, you suggested the possibility to simulate the whole human body development out of embryo... Anyway, my point is that it is not "understood" until proven. I do not care about ability to speak or human intelligence, I want to see any kind of proof, squid or worm nervous system will do, any traces of instinct or emotional response.

 

ANYTHING that can be considered evidence, you only point unsuccessful attempts. Can they simulate the simplest virus at least?

 

 

In other words, why do you hope Artificial Neural Network will be able to fully model neuron mechanics without considering the magnetic fields interaction? Do you think such AI based on deterministic algorithm could have a sense of humor, be able to dream, believe in god or commit suicide?

 

 

 

The great mysteries of the brain all involve its higher level structure and the interactions between various brain systems, not the behavior of individual neurons. That much is well understood.

 

 

Hold on, everything is well understood and deterministic... except that some high level interaction is a GREAT MYSTERY.

 

 

You can not claim thought process is deterministic, if the underlying interaction is MYSTERY.

You can not claim thought process is well understood, if the underlying interaction is MYSTERY.

 

 

You can not say anything about thought process if you do not know WHAT IT IS, WHERE IT IS, HOW IT ORIGINATES, HOW IT EXISTS, HOW IT EVOLVES and HOW IT CEASES TO EXIST. If you do not know mechanics of it, you can not simulate it. If you have no experiments to prove your theory then portraying it as a fact is scientifically unwise.

 

 

It is as if you want to simulate MS Word running inside computer, by looking at computer motherboard and the design of hardware, while being completely oblivious that computer needs to be turned on for hardware to be able to compute anything and display output. You can hardly reproduce software by looking at hardware. software is dynamical system, computer hardware is static. Do not get confused that human hardware is not static, information is still within electromagnetic fields, not within protein structure. Brain is Electrical Quantum Machine, not Mechanical Turing Machine.

 

 

YOU CAN NOT SIMULATE SYSTEM IF THERE IS ANY MYSTERY ABOUT IT'S MECHANICS.

 

This is my point from the start.

Now, realize your mistakes and accept this kindly given knowledge, do you accept?

Edited January 28, 2009 by Sione
Consecutive posts merged.

[bascule]

I thought, just before, you suggested the possibility to simulate the whole human body development out of embryo...

 

That would seem to be the "least common denominator" road to strong AI. Even if computer scientists and neurophysiologists are unable to crack the problem before then, I'm confident that computational biologists will end up picking up the slack.

 

Provided processing power continues to increase along present trends, I think a molecular simulation of an entire human being will be possible in about 50 years.

 

Anyway, my point is that it is not "understood" until proven.

 

Science isn't about "proof", it's about constructing the most logical argument based on the available evidence.

 

ANYTHING that can be considered evidence, you only point unsuccessful attempts. Can they simulate the simplest virus at least?

 

Present day attempts at building molecular models of cells are hindered by some particularly complex chemical reactions, such as the Gibbs free energy involved in protein folding. This has proven a tough nut to crack, although it is being researched simultaneously by the BlueGene/L supercomputer (which was the most powerful in the world) and the Folding@home project.

 

In other words, why do you hope Artificial Neural Network will be able to fully model neuron mechanics without considering the magnetic fields interaction?

 

Who's saying it won't? Certainly not me. That's a strawman. Again, the electromagnetic fields can be described using Maxwell's Equations, so what's the problem with incorporating them into the model? However, it's entirely possible that they aren't a necessary component of a successful simulation. If they are, they'll be included in the model.

 

Do you think such AI based on deterministic algorithm could have a sense of humor, be able to dream, believe in god or commit suicide?

 

It's a bit of a diction error to call something like a brain simulation an "algorithm". However, nitpicking aside, of course. The brain is, for all intents and purposes, a deterministic physical system. Or to put it in Tegmark's phraseology, "the degrees of freedom of the human brain that relate to cognitive processes should be thought of as a classical rather than quantum system"

 

I think what you're probably having trouble dealing with is the metaphysical separation between the physical systems responsible for consciousness and the first person conscious experience, which is clearly not "made of" anything physical.

 

I used to have trouble with this as well, then I started reading various interpretations of Kant, and began buying into the metaphysical separation between consciousness and the physical world.

 

So to go back to your original statement, I don't think the deterministic computer program itself "could have a sense of humor, be able to dream, believe in god or commit suicide?"

 

It is instead the conscious entity resulting from the operation of such a program which could have these experiences.

 

Hold on, everything is well understood and deterministic... except that some high level interaction is a GREAT MYSTERY.

 

Yes, while we have some pretty comprehensive theories for how the neocortex works, and understand the connection structure between the cortex and the thalamus, exactly how the thalamus and neocortex work in conjunction (through the interconnecting loops between them) remains a mystery.

 

You can not claim thought process is deterministic, if the underlying interaction is MYSTERY.

You can not claim thought process is well understood, if the underlying interaction is MYSTERY.

 

Yes, I can. What is not understood is how the high level structures work in conjunction. However, they are all made out of the same things: neurons and glia, and those are well understood.

 

We fully understand the building blocks and how they operate. There's nothing magical or quantum about them. They are merely physical systems.

 

YOU CAN NOT SIMULATE SYSTEM IF THERE IS ANY MYSTERY ABOUT IT'S MECHANICS.

 

Again, to reiterate, the low level mechanics are well understood. It's the high level structure that remains a mystery. Simulating complex systems is a great way to study them. Modeling has been used as a successful tool in a multitude of domains, particularly physics.

[Sione]

That would seem to be the "least common denominator" road to strong AI.

 

That is fine' date=' but we do not need to go that far.

 

We are not talking about intelligence or ability to speak, but emotions and feeling. What is your understanding on this, do following life forms have either emotion, instincts or both:

 

a.) Virus

b.) Bacteria

c.) Squid

d.) Bird

e.) Dog

f.) Dolphin

 

 

Science isn't about "proof", it's about constructing the most logical argument based on the available evidence.

 

Your statement is contradiction, evidence = proof.

 

In any case, your evidence points to failed experiments, which is neither evidence nor proof, it is more of rebuttal type of thing.

 

 

Present day attempts at building molecular models of cells are hindered by some particularly complex chemical reactions, such as the Gibbs free energy involved in protein folding. This has proven a tough nut to crack, although it is being researched simultaneously by the BlueGene/L supercomputer (which was the most powerful in the world) and the Folding@home project.

 

Is there anything left to argue then?

 

I can not write simulation of airplane based on opinion that lift is produced by Bernoulli's principle, if that is not real-world fact. I need to know real mechanics of flight, and ALL of it, if I am to ever be able to model it and get simulated plane of the ground, right?

 

 

...anyway, I'm glad we agree now.

 

 

 

Who's saying it won't? Certainly not me. That's a strawman.

 

I say, SPONTANEOUS actually means non-deterministic.

 

How do you simulate "spontaneous" with deterministic algorithms and hardware? Logically this seems impossible and my question is - do you accept random computer numbers as a source to provide true spontaneity?

 

 

 

Again, the electromagnetic fields can be described using Maxwell's Equations, so what's the problem with incorporating them into the model?

 

No, magnetic interaction can not be described with Maxwell equations and that is the problem with incorporating them into the model, which explains why no one has even tried it, right?

 

 

However, it's entirely possible that they aren't a necessary component of a successful simulation. If they are, they'll be included in the model.

 

What model?

 

You CAN NOT include magnetic interaction in Artificial Neural Network model, ANN is about floating point mathematics and magnetic interaction is about three-dimensional motion.

 

 

It's a bit of a diction error to call something like a brain simulation an "algorithm". However, nitpicking aside, of course.

 

By your understanding of the determinism of the system, word "algorithm" is not error, because you believe mathematically you can compute each next step based on current numbers and state of the system. That is exactly what algorithms do and you actually suggest it is possible to simulate consciousness with mathematical algorithms.

 

I should be the one to complain about it, it is me who says algorithms can not do it because they are inherently deterministic.

 

 

The brain is, for all intents and purposes, a deterministic physical system.

 

Will you stop contradicting yourself, you already admitted brain is a GREAT MYSTERY.

 

 

So to go back to your original statement, I don't think the deterministic computer program itself "could have a sense of humor, be able to dream, believe in god or commit suicide?"

 

So, now we agree?

 

What is it your are saying? Is brain deterministic or not?

If yes, then why could not deterministic program produce the same stuff as deterministic brain?

 

This is exactly what you were suggesting so far, brace yourself and pick your stance, you can not agree and disagree in the same time.

 

 

You can not claim thought process is deterministic' date=' if the underlying interaction is MYSTERY.

You can not claim thought process is well understood, if the underlying interaction is MYSTERY.

[/quote']

 

Yes, I can. What is not understood is how the high level structures work in conjunction. However, they are all made out of the same things: neurons and glia, and those are well understood.

 

We fully understand the building blocks and how they operate. There's nothing magical or quantum about them. They are merely physical systems.

 

You can, but it is nonsense.

 

I'm talking about dynamical PROCESS and you are talking about building blocks. Knowing the shapes of gears in a clock does not translate to knowledge about building machine that measures the time, especially if you fail to realize the clock you are trying to re-create is not spring based, but electrical.

 

You do not know what thought process is, no human does.

 

 

Do you think chemistry is deterministic?

This should not be too hard to prove or disprove, eh?

 

 

YOU CAN NOT SIMULATE SYSTEM IF THERE IS ANY MYSTERY ABOUT IT'S MECHANICS.

Again' date=' to reiterate, the low level mechanics are well understood. It's the high level structure that remains a mystery.

[/quote']

 

Do you agree or not?

Your response is not related to what you quoted, I'm talking about mechanics and you are talking about structures.

 

 

High level, low level... I do not care what you imagine to know, it is not complete knowledge and therefore you can not know how much you do not know. Even if there is 0.0001% MYSTERY, you DO NOT KNOW if the importance of it to achieve true simulation could be 99%. It is nonsense to make any statement about certainty of system dynamics and in the same time acknowledge great confusion about it.

 

 

Could you finally explain why do you hope incomplete simulation will produce results?

Edited January 29, 2009 by Sione