When will machines be as smart as humans?

[Complexity]

A regular computer can do anything a neural net can do. Once your computer is Turing-complete, architectural specifics make no difference. That's what the Church-Turing thesis is all about.

PixyMisa is absolutely correct.

 

[Complexity]

I think that what we currently know about classical physics can't explain consciousness because people are still arguing about it. If it could explain consciousness, it would have, and things like this thread would not go on for as many pages as it has..

This has got to be one of the most idiotic statements I've read in years, and that is saying a lot.

 

[a_unique_person]

Not quite. We can't explain it yet. However, that doesn't mean we won't, one day. It won't be physics that explains it, however, but advances in logic, intelligence, etc.

 

[Belz...]

PixyMisa is absolutely correct.

Does that mean that computers are already sentient ? I mean, some already have more computing power than some animals that are, probably, at least partly self-aware.

 

[Soapy Sam]

I don't know why you believe that biological systems are so special compared to non-biological ones. Both are, quite simply, composed of the same basic stuff.

The same basic stuff, yes. But the process of creation is utterly different.

An apple tree and an apple pie contain the same basic stuff, but are hugely different. We can make apple pies. The only way we can make apple trees is by using the existing system of self assembly. Which is more complex - a tree, or a conscious mind?

Neural organisation is the result of 3 billion years of evolution which has tweaked the interdependancy of the cells, of the DNA in the cells, the environment in which they develop (a womb), the chemistry of the mother's body, the presence of non nuclear DNA (as in mitochondria) and so on and on, in a massively re-entrant cascade of self referential processes, which (these days) takes around 40 weeks in humans. The result is a thing primed to learn, programmed to acquire language, and aware of it's own existence, which then interacts with an environment that stuffs it with information for the next two decades. This is not a model of anything. It's a real baby. Is this our experience of computers?

Its a simple fact that only living systems display consciousness. It is that fact alone which leads me to think it probable that consciousness actually is a result of being alive. I postulate no mystical forces, merely that an object must have undergone a process akin to embryonic development before it can become conscious, because that is where whatever it is that consciousness is actually happens. If that process can be copied, then the resulting thing might indeed be conscious, but that process is not how computers are built.
This is why I think Pixy Misa is wrong to place his faith in computers capable of modelling brains producing the behaviour of brains. I see no reason to think that is true. It's an assumption, as much as my assumption that the result depends on the process. Modelling is not enough. The synthetic mind must actually be a mind, not a model. I think the mind gets started as part of the development process, so reproducing the end architecture simply is not enough. It's cargo cult engineering.

So I think we must grow a brain, the parts of which are aware from the inception . That may be possible synthetically, but it's not what we are doing now. Some form of nanoengineering might be a better route to take, where the logic elements self- assemble into a final structure, but still I think there are many processes we are unaware of, any of which could be responsible for consciousness. (But none of them necessarily requires quantum anything. They might, but we have no reason to suppose so.)

So in summary. I think the computer route can probably build machines which behave so intelligently that we would have difficulty telling their intelligence from human intelligence. I think these would be the famous p-zombies. Their substrate might be aware, but the whole thing would not be self aware- conscious.

I think to create true consciousness we must learn to copy not merely the material end product of embryonic development, but the entire process, or we risk missing many possibly critical steps.

I do not think we are likely to be capable of that for decades.

No reason not to start though.

But let's face it. We do not know what consciousness is, though we know some of what must be involved. We are all arguing here from a position of some ignorance. In reality all we can do is build stuff and see what happens.

 

[Soapy Sam]

This has got to be one of the most idiotic statements I've read in years, and that is saying a lot.

Complexity- Does this mean you think classical physics actually has explained consciousness? If so, can you point me to a link, or explain it here? I'm not sure what you find so idiotic about Beleth's remark.

 

[elliotfc]

Oh, and also that (a) Xeno's Paradox was solved hundreds of years ago and (b) if you've got half-way there, you can just buy two computers.

I was being fuh-see-shus.

-Elliot

 

[Belz...]

The same basic stuff, yes. But the process of creation is utterly different.

Still, there's no reason to believe that you can't get the same result from an artificial mind.

For example, I'm sure you can imagine an artificial arm that duplicates every movement and sensation that you'd expect from a normal arm. Why would this be different with the brain ?

Which is more complex - a tree, or a conscious mind?

Haven't a clue, really.

Neural organisation is the result of 3 billion years of evolution which has tweaked the interdependancy of the cells, of the DNA in the cells, the environment in which they develop (a womb), the chemistry of the mother's body, the presence of non nuclear DNA (as in mitochondria) and so on and on, in a massively re-entrant cascade of self referential processes, which (these days) takes around 40 weeks in humans. The result is a thing primed to learn, programmed to acquire language, and aware of it's own existence, which then interacts with an environment that stuffs it with information for the next two decades.

Irrelevant. It took just as much time to design the human arm. See my example above.

This is not a model of anything. It's a real baby. Is this our experience of computers?

Again, this is irrelevant. Who cares HOW much time it takes to get to the point we're talking about, or HOW it does this ? The point is, if the process is the same and we can replicate whatever causes consciousness is biological organisms, we should get the same results, even if it only takes you a week to figure it out.

Its a simple fact that only living systems display consciousness.

They are also the only ones to display controlled flight, but we all know that changed once we understood how that worked.

It is that fact alone which leads me to think it probable that consciousness actually is a result of being alive.

Really ? I'm not sure I follow. Single-cell organisms are not aware, and there's surely a good deal of complex animals and plants (mostly plants) that DON'T have consciousness. It's likely that consciousness arises from a certain electric field configuration, which is NOT DEPENDENT upon the composition of the emitter itself.

I postulate no mystical forces, merely that an object must have undergone a process akin to embryonic development before it can become conscious, because that is where whatever it is that consciousness is actually happens.

This is completely baseless. Please provide SOME evidence to back this up. Otherwise I'll come up with a number of crazy theories, too.

If that process can be copied, then the resulting thing might indeed be conscious, but that process is not how computers are built.

So... are you saying that... if we construct a clone of someone, grown in a vat, and we rearrange his neural net to correspond to that of the original, he won't have the same (read: not shared) basic consciousness ?

Modelling is not enough. The synthetic mind must actually be a mind, not a model. I think the mind gets started as part of the development process, so reproducing the end architecture simply is not enough. It's cargo cult engineering.

Huh ? Who's talking about a model ? A computer is a real thing and it's processing is real, too. If, per my assumption above, its "neural" configuration gives rise to an electric field similar to that of the human brain, it's likely that it'll have a mind, no matter what your definition of that word might be.

 

[Jorghnassen]

I don't know why you believe that biological systems are so special compared to non-biological ones. Both are, quite simply, composed of the same basic stuff.

Well, my guess is because, as much as we'd love to be able to reduce biology to chemistry then to physics, we just can't do it (at the moment). Take two living identical twins. Suffocate one to death. Both twins are still composed of the same basic stuff. But the dead one just can't do anything (by itself) anymore. Similarly, all atoms are made of the same basic stuff (protons, electrons and neutrons, well except for hydrogren that isn't deuterium or tritium), but different elements have different properties. And the same goes with chemical compounds. Sodium by itself, highly reactive. Chlorine by itself, is poisonous to living creatures. Put them together and you can stick the result on your steak and eat it.

So conciousness/intelligence might be achievable without living parts involved, but there's no way to tell for now. So, until our understanding of either intelligence, conciousness, the brain or computers advances to the point where we can have an answer for this question (or avoid it by making some sort of biological artificial machine which often appear in sci-fi stories), we can't handwave it with a reductionist argument, or do like PixyMisa and say a Turing machine can do it (Turing machines are nice theoretical entities and thought experiments, but they practically exist only in the platonic realm of ideas...).

 

[cyborg]

The same basic stuff, yes. But the process of creation is utterly different.

I fail to see why the process is important. It is the structure that leads to the function, not the way that structure came to exist.

Its a simple fact that only living systems display consciousness.

Well it ain't that simple because there's no robust way of determining if something is concious other than by the opinion of something else that's concious.

So how would you determine if a non-living system was concious or not?

It is that fact alone which leads me to think it probable that consciousness actually is a result of being alive. I postulate no mystical forces, merely that an object must have undergone a process akin to embryonic development before it can become conscious, because that is where whatever it is that consciousness is actually happens. If that process can be copied, then the resulting thing might indeed be conscious, but that process is not how computers are built.

Yes but being alive isn't anything special on the chemical level. What separates life from death is the fact that the chemical systems have stopped reacting in an 'alive' way and without those systems in place the chemical systems start to fail until it is basically impossible to revert to the previous state.

If it were possible to manipulate that chemical failiure such that normal 'alive' chemistry were to resume then clearly it would be possible to construct an entire system in such a way.

I cannot see why you think the embryonic process is so special. In my mind it is merely the evolutionary selected mechanism by which a new life system can be constructed. I think it is easy to imagine that such a process coming into existence to build multicellular organisms is not that surprising because alternatives would be harder to effect.

That is to say it's really hard to build a fully formed lifeform and then 'switch' it on precisely because it's a really hard problem. Some things are just hard to do - like balancing a million marbles on your nose.

So it all boils down to a simple question really: in what way must there be a special physical property that can ONLY be attained by the embryonic process and absolutely cannot be produced by any other imaginable physical means?

The synthetic mind must actually be a mind, not a model. I think the mind gets started as part of the development process, so reproducing the end architecture simply is not enough.

So again it would be impossible to model this right?

Why?

And why can't I do this just once and copy the brain? Do I HAVE to grow it each time?

Why?

I think these would be the famous p-zombies. Their substrate might be aware, but the whole thing would not be self aware- conscious.

And if they considered you to be a p-zombie?

I think to create true consciousness we must learn to copy not merely the material end product of embryonic development, but the entire process, or we risk missing many possibly critical steps.

So the begging question is:

I have an artificially produced brain by any mechanism whatsoever you can deem acceptable. I know all its inputs and all its outputs for all possible data. (Yes impractable but bear with me). Why is the machine concious but not the look-up-table I've constructed that produces identical results?

Can you give me any evidence that conciousness is a quality that is process derrived only?

It just seems to me that you're denying the only sensible conclusion: conciousness is a reciprocal property, not a physical one.

 

[Belz...]

So conciousness/intelligence might be achievable without living parts involved, but there's no way to tell for now.

Perhaps, but it seems much more likely that it's more a matter of configuration or computing power than composition. After all, speakers can emit sounds but they sure as hell aren't made like larynxes and lungs.

 

[Jorghnassen]

Perhaps, but it seems much more likely that it's more a matter of configuration or computing power than composition. After all, speakers can emit sounds but they sure as hell aren't made like larynxes and lungs.

Granted. I'm just the type to warn against reductionism, too much optimism and paying only attention to theory over application to reality. In theory, flying cars are possible. In practice, it's not so easy...

 

[Belz...]

Granted. I'm just the type to warn against reductionism, too much optimism and paying only attention to theory over application to reality. In theory, flying cars are possible. In practice, it's not so easy...

Actually, it is. Just REALLY expensive.

 

[cyborg]

Hell, I saw a TV program where this guy actually tried to build a flying car. Not sure if he ever got it off the ground.

 

[Soapy Sam]

Belz-
Still, there's no reason to believe that you can't get the same result from an artificial mind.

Perhaps. On the other hand , I see no reason to assume I could. We are just exchanging opinions.

For example, I'm sure you can imagine an artificial arm that duplicates every movement and sensation that you'd expect from a normal arm. Why would this be different with the brain ?
I can imagine it. Can you build it? Every movement and sensation? I don't think that can be done, no.

Irrelevant. It took just as much time to design the human arm. See my example above.
"Design" ? Arms lift, push and pull. Brains not only control arms, they do a great deal more. I think there is a big difference. And I do not believe we can reproduce a human arm, except by growing one.

Again, this is irrelevant. Who cares HOW much time it takes to get to the point we're talking about, or HOW it does this ? The point is, if the process is the same and we can replicate whatever causes consciousness is biological organisms, we should get the same results, even if it only takes you a week to figure it out. That's verging on tautology. Obviously if we can reproduce whatever causes x we can produce x. That's precisely my point- we don't know what produces consciousness and by assuming it's purely neural architecture we run the risk of missing all the developmental process which may be critical.

We expect our machines to be built and then turned out as finished products. I can only repeat - this is not how organisms work. They are functional from the moment of fertilisation and timing is critical in development. You feel that's irrelevant. I do not.

They are also the only ones to display controlled flight, but we all know that changed once we understood how that worked. Yes, but I have yet to see something which displays controlled flight and is neither alive nor designed by a biological intelligence. Can you think of one?

I'm not sure I follow. Single-cell organisms are not aware, and there's surely a good deal of complex animals and plants (mostly plants) that DON'T have consciousness. It's likely that consciousness arises from a certain electric field configuration, which is NOT DEPENDENT upon the composition of the emitter itself. Really? What makes you think this is likely? You may be right, but I doubt you can prove it . You are arguing your bias, like me. (Which is fine). I never said all living things are conscious. I said all conscious things are living. That may be coincidence. I'm sceptical.

This is completely baseless. Please provide SOME evidence to back this up. Otherwise I'll come up with a number of crazy theories, too.
My observation is simplistic- Every conscious entity, without exception has undergone a complicated process of development. You see this as irrelevant, I think it's fundamental. We'll agree to differ. I do not think it's a crazy theory though.

So... are you saying that... if we construct a clone of someone, grown in a vat, and we rearrange his neural net to correspond to that of the original, he won't have the same (read: not shared) basic consciousness ? I'm not sure what you mean by "construct". If we grow a human clone as we grew Dolly the sheep , I'd expect him to be mentally very like the person the original cell came from, once we correct for nurture, education, embryonic chemistry etc. I don't understand your reference to a neural net. If you mean literally " construct" then I have no idea what we would get, or where we would start, and I don't think anyone else has either.

Huh ? Who's talking about a model ? A computer is a real thing and it's processing is real, too. If, per my assumption above, its "neural" configuration gives rise to an electric field similar to that of the human brain, it's likely that it'll have a mind, no matter what your definition of that word might be.
The whole thread is about computers modelling human consciousness. I think your reasoning here is circular: If mind is an electric field and two fields are the same, then there are two minds. But is mind an electric field? Could be. If human minds are electric fields generated by brains and the structure of brains is a result of embryonic development, will different development produce the same field? It might. It might not. I don't think we know. I suspect it would not.

 

[Soapy Sam]

I fail to see why the process is important. It is the structure that leads to the function, not the way that structure came to exist.

You are absolutely sure of that? I'm not. Imagine a production line had to build a car in such a way that at every moment in the process of assembly, the car was actually functional and functioning. (Not necessarily as a car).
Would the process affect the end result? Would there ever be an end result? At what point is it "finished" if it is permanently functioning? That is how organisms are put together- they are alive all the way. I see this as radically different from machine assembly and I expect the difference to be reflected in the product.


Well it ain't that simple because there's no robust way of determining if something is concious other than by the opinion of something else that's concious.
True. Can you think of a single example of something you, as a conscious entity think is conscious and which is not organic? I can't. Hence my observation.

So how would you determine if a non-living system was concious or not?
In the same way I would determine if a living system was conscious- by studying its behaviour.



Yes but being alive isn't anything special on the chemical level.
Depends what you mean by "special". If you are accusing me of being " livingist", I plead guilty as charged. I think there is something very special about life at the chemical level. Not the chemicals themselves, obviously, but the process by which they interact. It's a matter of degree though, not of kind. As I have said several times, I do not believe in elan vital.
What separates life from death is the fact that the chemical systems have stopped reacting in an 'alive' way and without those systems in place the chemical systems start to fail until it is basically impossible to revert to the previous state.

If it were possible to manipulate that chemical failiure such that normal 'alive' chemistry were to resume then clearly it would be possible to construct an entire system in such a way. You've been reading Mary Shelley?
I cannot see why you think the embryonic process is so special. In my mind it is merely the evolutionary selected mechanism by which a new life system can be constructed. I think it is easy to imagine that such a process coming into existence to build multicellular organisms is not that surprising because alternatives would be harder to effect. Merely?

That is to say it's really hard to build a fully formed lifeform and then 'switch' it on precisely because it's a really hard problem. Some things are just hard to do - like balancing a million marbles on your nose. Yup.

So it all boils down to a simple question really: in what way must there be a special physical property that can ONLY be attained by the embryonic process and absolutely cannot be produced by any other imaginable physical means? I'm sure I mentioned no such physical property. I agree absolutely that evolution builds critters this way because it's the only way it can be done. The only thing that is special (in the chemical sense you mentioned before)- is process. The thing is alive from the start, unlike a car or a computer. It operates on itself, from the start. It modifies it's own development in response to external and internal events. It produces structures of astounding regularity and complexity, which are not only viable , but functioning at every instant of the process. The process does not stop at birth. It stops at death. Somewhere along the line, this thing becomes self aware. Somewhere along the line, it may cease to be.
You seem to think a baby could be (in principle) assembled dead, from off the peg parts and kick started at the moment of birth and that a normal human intelligence would result a la Frankenstein.

I think that's very , very improbable.



So again it would be impossible to model this right? "Impossible"? I don't know. I'm pretty sure we can't do it with computer technology. I think nanotech might be a way, but let's face it - most of what we "know " of that is fiction.

Why? I can't think of existing technology which can produce life, far less conscious life. Can you?

And why can't I do this just once and copy the brain? Do I HAVE to grow it each time? Yes. For the reason given above. For a fully formed brain to be conscious , an earlier part of the continuous process has to operate at the correct time. You can't just play a mind tape into a piece of finished machinery, any more than you can blow life into it.

Why?



And if they considered you to be a p-zombie? If who did? And if so, what? I don't get your meaning here.



So the begging question is:

I have an artificially produced brain by any mechanism whatsoever you can deem acceptable. I know all its inputs and all its outputs for all possible data. (Yes impractable but bear with me). Why is the machine concious but not the look-up-table I've constructed that produces identical results? The bus arriving in London from Glasgow has no passengers from Cardiff. It did not go through Cardiff en route. Why is this so hard to see?

Can you give me any evidence that conciousness is a quality that is process derrived only? Yes. The process is chemical and is generally called "life". Nothing not alive displays consciousness- even when examined by other conscious entities. The evidence for this is all over the place.

It just seems to me that you're denying the only sensible conclusion: conciousness is a reciprocal property, not a physical one.

Sorry- I don't understand what you mean here by "reciprocal". For me, consciousness is a chemical process characteristic of some living systems. I have seen nothing resembling it in non living systems. Intelligent behaviour is a far simpler concept. I have no doubt at all that intelligent behaviour can be emulated by machinery. I do doubt such machinery will ever be conscious, though radically different machinery may be some day. I just think if that's what we are after, computers are not the way to go.

 

[Roboramma]

Soapy Sam, just a little clarification:

Imagine that it was possible to "build" a human body in a way different from the way they traditionally come to be. I'm imagining some sort of nanotechnology putting it together much like we put together traditional machines.
The final product comes out physically identical to me, for instance. Would it be concious? If there is no soul, if there is nothing that is not material, then how can I and this construct be different, if we are physically identical?

Is your argument that doing so would be impossible? I won't say it is, but if that's what you're saying I think you are arguing at cross purposes with Belz. He is saying if we could build something that was physically identical to a human mind in all ways that matter for cognition, then it would be concious.

You are saying that the process of building that mind is important, but isn't it only important in how it effects the outcome?

 

[Soapy Sam]

Soapy Sam, just a little clarification:- Roborama

Imagine that it was possible to "build" a human body in a way different from the way they traditionally come to be. I'm imagining some sort of nanotechnology putting it together much like we put together traditional machines.

Dunno. Nano seems like a possibility as I said above, but it's mostly still SF.

The final product comes out physically identical to me, for instance. Would it be concious? I don't know. The assumption of the Frankenstein school is that it must, because the final hardware is the same. My assumption is that the process of life (which I see as purely chemical, but hellishly complex) is continuous from the start . At no time during construction is the thing not alive. Consciousness develops during that process.
-A simplistic analogy. I build a house. Before putting in the internal walls and floor, I lay the electrical wiring. On the day I sell you the house, you walk in and switch on the power. But what if there is no power switch? What if the power has to be turned on while I was building the house?

That's what I mean about the history of the process being critical. If I failed to connect the power before I put the internal walls in, you are screwed.It's too late to do it now.
If there is no soul, if there is nothing that is not material, then how can I and this construct be different, if we are physically identical?
No soul. The only non-material thing is process. A car rolls off a production line. It's an inert amalgamation of metal and plastic. Another car rolls off another line, but it's engine is running, the fuel tank and wash bottles are full , it's lights are on and the radio is playing- and they have been since various different stages along the production line.

Would you say these two cars are identical? Would you say the process of manufacture was identical? Would you say either car requires a soul to explain it's behaviour?

Is your argument that doing so would be impossible? I won't say it is, but if that's what you're saying I think you are arguing at cross purposes with Belz. He is saying if we could build something that was physically identical to a human mind in all ways that matter for cognition, then it would be concious.
I don't know if we could. I can't think how, except by the time honoured method. My argument is that the architecture is not enough. It has to be operating. I think consciousness is a property of living things (I await a counter example) and that living things are all assembled in the way I have described. I do not think there is any proof that a "built" brain would be conscious, or that a built body would be alive. This is the Frankenstein assumption. I think, for some reason, many people here are making that assumption without realising that it is an assumption.

You are saying that the process of building that mind is important, but isn't it only important in how it effects the outcome?
I don't think so, no. The end product has to be alive. And it has to be alive at every stage of development. There is no way to switch it on after it is built. That's all I'm saying. It's astonishingly simple.
I'm honestly puzzled that people seem to have difficulty with this. There is nothing "woo" in this, no spirit, or vital essence, merely the assumption that what I see in nature is reality- that all conscious things are living things and that all living things are grown- and are living at every point of the growth process.

I'll repeat that I'm talking about consciousness. Intelligence I see as something simpler. We judge intelligence by behaviour and I have no quibble with present day computers displaying intelligent behaviour. Searle's Chinese room displays intelligent behaviour, but the only thing in the room which is conscious is the man.

 

[Beleth]

This has got to be one of the most idiotic statements I've read in years, and that is saying a lot.

You know, I have a pretty good imagination.

I've helped design role-playing games. I've built computer networks that became industry standards five years later. I've made up impromptu half-hour-long children's stories that those same children ask me to repeat for years.

But not even when my imagination is most vivid can I come up with a situation where I would possibly care in the slightest what you thought of me, sir.

Now if you'll excuse me, and even if you won't, I need to construct a response to one of the intelligent posters around here.

 

[Beleth]

Not really. It's all QM.

I know you know better than that, because you said it before yourself. QM and relativity are irreconcilable.

That doesn't make any sense. What extension? How? Where? Why?

That's just it. If it's an unknown branch of physics, or unknown properties of an existing branch, then by definition I can't answer those questions, and neither can anyone else.

There simply isn't anything going on in the brain that isn't explained by existing physics.

Then I'm sure you will find it a trivial task to produce the equation that produces consciousness.

Have you ever seen an Ian thread? Or a Kumar thread? Or those threads on raging kundalini or whatever it was? Von Neumann's incoherent attacks on evolution?

I'm not just talking about the nutjobs here. (So yes, my example wasn't the best.) I'm talking about smart people, in academia, still discussing it. People like Penrose.

Just because someone argues about it doesn't make much of a case.

It means it's not truly settled. Learned people don't argue whether gravity exists, or its properties, any more. And they don't argue over whether the world is round or not. But they still argue over the nature of consciousness. When people who are smarter than I am in a subject still don't agree, well, that gives me pause.

I'm not asking you to. I'm asking you to present some evidence - anything at all - that actually suggests that something in the brain or in consciousness is not explicable by classical physics.

I hope my clarifications above have done that.

Yes, I am appealing to authority. But that's not a fallacy when the authority appealed to is an actual authority on the subject matter being discussed.

Consciousness is generated by the brain. There is simply too much evidence to rationally doubt this.

The brain is a biological system.

Biological systems are entirely explicable in terms of chemistry.

Chemical systems are entirely explicable in terms of physics.

QED.

Ahhh, but there's a huge difference between "explicable" and "explained".

I agree that consciousness is explicable by physics.
I do not agree that our present knowledge of physics can explain it.