Should AI be regarded as Turing-test verified?

[Kyriakos]

The question is why should that minuscule inexactness matter? I can't calculate the behavior of all the electrons in my pocket calculator but I can still say exactly what it reads out when I enter a given input. That's where I think you're hiding magical thinking. What is it about neurons that makes them able to bring about intelligence that you can't do with electronics?

We are talking about 'replication'. Last time i checked that means something other than approximation up to something Let alone that it is not really likely that biological material can be approximated and still function with no evident differences (which for a bio form can spell horrible issues, given that even smallish changes in the (eg) human genome lead to things like elephantiasis etc).

Your "magical" line is just projection, though. I would not be claiming that if you had an actual replication of X then the actual replication would not be X due to "some magic gone" I argue why you likely cannot have an actual replica in the first place. Different kind of lost magic..

 

[Zelig]

No actual sphere exists 'in reality' anyway.

So what? You can still 3d model any physical object.

 

[Perfection]

All replicas are mere approximations. The question is if it's good enough. The magical thinking is in saying that because no replica is perfect it must be missing something important. What important property of an ant cannot be replicated in robo-ant?

 

[Kyriakos]

All replicas are mere approximations. The question is if it's good enough. The magical thinking is in saying that because no replica is perfect it must be missing something important. What important property of an ant cannot be replicated in robo-ant?

I am just asking that the actual ant is fully replicated. Y U NO fully replicate what i am typing for the last couple of pages?
3,140 is not pi, sorry

 

[Perfection]

I get that, I'm asking if the replica isn't perfect what important difference is left out. What about that makes robo-ant so unlike the regular ant? Why does the difference matter?

 

[Kyriakos]

I get that, I'm asking if the replica isn't perfect what important difference is left out. What about that makes robo-ant so unlike the regular ant? Why does the difference matter?

Already used a parallel? (3,14 not being &#960.
I see no reason to believe that an imperfect replica of biological matter should be deemed as that biological matter it is an imperfect replica of. You with elephantiasis would virtually have your current biological make-up. Yet surely the difference would be rather evident.

 

[dusters]

wait, since there is always sex involved, wasn't it the case that the cloned sheep were always sterile?

and your robot-ant, it wouldn't be able to procreate with other robot-ants, right?

edit:

just checked - the cloned animals are able to give birth, so it must not be the case. However, i do find life to be something of magical quality itself. Cloning so far has not been very succeful, the scientists have cloned a lot of mammals, but with low chance of animal surviving.

so this AI problem, if you can create an AI which can create another, even dumb AI, that might be the final solution?

 

[Perfection]

Worker ants are sterile, but procreation possibility shouldn't matter or people who can't reproduce wouldn't be intelligent.

 

[Kyriakos]

Worker ants are sterile, but procreation possibility shouldn't matter or people who can't reproduce wouldn't be intelligent.

Having absolutely zero ability to reproduce likely would spiral any human intelligence down rapidly, cause a significant percentage of the human body functions as a means for said human to be granted (either more directly, easily, or maybe even uniquely) needed positive/pleasurable emotions by making you try to have those kinds of relationships.
Again, less ability is not the same as zero ability. Another facet of why replication is not about mimicking something to a fixed degree, at least in the matters discussed here

 

[dusters]

i'm going at the fact that people haven't invented anything self-sustainable, no new species, which can survive without further interference?

and this AI, this AI has to be so smart to find a way to self-sustain itself completely, it needs survival instinct. How do you programm an instinct?

 

[El_Machinae]

All replicas are mere approximations. The question is if it's good enough. The magical thinking is in saying that because no replica is perfect it must be missing something important. What important property of an ant cannot be replicated in robo-ant?

Well, with an ant, it's a bit of a different story compared to sentience. Now, all animal behaviour can be mimicked or replicated using information theory. So, with an ant, there's no real difference. Bio-ant will act like robo-ant.

But with sentience, it's possible to replicate its appearance. The problem is that we don't know what confluence of physical forces are necessary for sentience. Does it require information processing at the level we see with neurotransmitters (which is quantized at the vesicle level)? Does it require a specific overlap of EM fields? We just don't know.

The one insight might be to acknowledge the uniqueness of qualia. A qualia contains a great deal of information about the environment (i.e., 'red' gives you insight into the energy of the photons hitting your retina), but it also contains an informational je ne sais quoi that we just cannot seem to be put into informational form. How many bits are required to get you to imagine the same shade of red I'm currently imagining? We don't really know, but it's a lot.

Does the sentience organism therefore de facto contain more information than we think should be achievable through simple programming? So, an AI passing the Turing Test might be conversant, but (if it lacked sentience) cannot access the extra dimension of information contained within a qualia.

That said, our sentience is so utterly coarse compared to how much processing is going on inside our brains. It could be that sentience doesn't give any additional insight.

 

[Kyriakos]

Well, with an ant, it's a bit of a different story compared to sentience. Now, all animal behaviour can be mimicked or replicated using information theory. So, with an ant, there's no real difference. Bio-ant will act like robo-ant.

But with sentience, it's possible to replicate its appearance. The problem is that we don't know what confluence of physical forces are necessary for sentience. Does it require information processing at the level we see with neurotransmitters (which is quantized at the vesicle level)? Does it require a specific overlap of EM fields? We just don't know.

The one insight might be to acknowledge the uniqueness of qualia. A qualia contains a great deal of information about the environment (i.e., 'red' gives you insight into the energy of the photons hitting your retina), but it also contains an informational je ne sais quoi that we just cannot seem to be put into informational form. How many bits are required to get you to imagine the same shade of red I'm currently imagining? We don't really know, but it's a lot.

Does the sentience organism therefore de facto contain more information than we think should be achievable through simple programming? So, an AI passing the Turing Test might be conversant, but (if it lacked sentience) cannot access the extra dimension of information contained within a qualia.

That said, our sentience is so utterly coarse compared to how much processing is going on inside our brains. It could be that sentience doesn't give any additional insight.

Re your example with bits and imagining specific shades of red:

Factored in should be that you are not asking (for example) three humans to place themselves at X,Y,Z angles to each other, but asking 3 or 4 or 2 or 5 and fractions etc, humans to place themselves at the "same" position isomorphically. The result (even if we suppose could be the "same") is not going to be developed from the same procedure, due to the variables moving to get the result being different.
Apart from all of us being (likely ) human, we all have a unique human set-up of mental world/existence.

 

[Perfection]

Already used a parallel? (3,14 not being &#960.
I see no reason to believe that an imperfect replica of biological matter should be deemed as that biological matter it is an imperfect replica of. You with elephantiasis would virtually have your current biological make-up. Yet surely the difference would be rather evident.

The difference is evident, but the relevance is questionable. A metal replica of a wood chair is also a chair even if the replica is imperfect. As far as chairness is concerned it doesn't matter if it's made of biological material or inorganic. When it comes to intelligence, people often say biological material does matter. When pressed as to why should that be they don't seem to have a straight answer or they have one based on magical thinking (some special life force). I'm saying it doesn't matter because there's no reason to suppose meat is any more special then the junk computers are made of.

 

[Kyriakos]

"Special life force" is to "unknown paragons by parts subatomic to your cut" as "lighting is the weapon of gods" to "lighting is a phenomenon to study". Ie very tenuous relation, and usually (by now) comes with an undercurrent of not wanting to see what the other person is saying?

 

[Perfection]

Why does that subatomic bit matter? I'm not saying there isn't a difference. I'm asking why it's important. Can you answer that? In the case of chairs organcness doesn't matter. Why should it for intelligence?

 

[Kyriakos]

^You know that it is not currently possible to answer that (i mean the "why" not the guess that it does matter if you aim to actually replicate something you don't know 'fully')) with specifics, cause we just don't know how biological matter is affected by yet smaller parts not examined yet. As for stuff like stones etc, well at least there there isn't much at stake if you make a replica of a stone. It is not like it is prone to develop some tumors and finally breed sinister and unusual worms or whatever

Why don't you look at it from the physics/quantum pov, and accept that parts below the known ones can indeed cause us to crudely round up the theory/examination of what we can already observe? Surely no one in physics would claim that given we already reached a level of examining particles, there is no reason to seek further depth if we aim to actually "know" what rules govern matter in those levels, linking it to laws we observe for larger levels?

 

[Perfection]

Why don't you look at it from the physics/quantum pov, and accept that parts below the known ones can indeed cause us to crudely round up the theory/examination of what we can already observe? Surely no one in physics would claim that given we already reached a level of examining particles, there is no reason to seek further depth if we aim to actually "know" what rules govern matter in those levels, linking it to laws we observe for larger levels?

We may not know the real low level quantum phenomena, but we don't have to. The kinds of phenomena that we don't know have extremely high or low energy levels. They're not the kinds of phenomena that could effect the way the brain works. The secret to consciousness doesn't lie in these arcane low level laws. It's like you're looking for things to move a boulder in your garden, and you have an inventory of everything bigger then a grain of rice and smaller than the moon. Rice sized things and moon sized things are improperly sized to be useful. (one might bring up the idea of a pile of rice, but rice piles are bigger then a grain of rice and would be included in said inventory).

One thing to ask yourself is what kind of rules about physics would matter for the purposes of saying neurons are so different then transistors?


Another thing to consider is the Copernican principle. It used to be thought that Earth was special and the center of the universe, but we now know that to be false. It used to be thought that there was only the milky way galaxy, but we also now know that to be false. It used to be thought that organic chemicals couldn't be produced from inorganic chemicals, but we now that to be false. People nowadays think brains are somehow made of special stuff with unique physical rules that separate them from inferior computers. The Copernican bet is to say, no, brains don't follow special rules.

 

[Kyriakos]

We may not know the real low level quantum phenomena, but we don't have to. The kinds of phenomena that we don't know have extremely high or low energy levels. They're not the kinds of phenomena that could effect the way the brain works. The secret to consciousness doesn't lie in these arcane low level laws. It's like you're looking for things to move a boulder in your garden, and you have an inventory of everything bigger then a grain of rice and smaller than the moon. Rice sized things and moon sized things are improperly sized to be useful. (one might bring up the idea of a pile of rice, but rice piles are bigger then a grain of rice and would be included in said inventory).

One thing to ask yourself is what kind of rules about physics would matter for the purposes of saying neurons are so different then transistors?


Another thing to consider is the Copernican principle. It used to be thought that Earth was special and the center of the universe, but we now know that to be false. It used to be thought that there was only the milky way galaxy, but we also now know that to be false. It used to be thought that organic chemicals couldn't be produced from inorganic chemicals, but we now that to be false. People nowadays think brains are somehow made of special stuff with unique physical rules that separate them from inferior computers. The Copernican bet is to say, no, brains don't follow special rules.

And more aptly/notably:

We may not know the real low level quantum phenomena, but we don't have to. The kinds of phenomena that we don't know have extremely high or low energy levels. They're not the kinds of phenomena that could effect the way the brain works.

The above does not seem to be worth being written categorically. Surely no one currently can make such a claim and sustain it with actual proof. And while your example included some difference in level of planes (moving some material we see, using the moon etc, or the opposite, using microscopic changes) you surely have to note that even moving from atoms to molecules the difference in scope and law-examination is simply massive. Enough for the former to be in Chemistry, and the latter in particle Physics. Just because we have not yet discovered the analogous next level "nucleus erosion" leading us to examine sub-atomic parts for the nuclear phenomena, doesn't at all mean that there is nothing like that there. It is not as if we can claim that the divisions of matter end somewhere, and even if they did we have no reason to claim they end at the current particles nor that events below those play no actual role in the previous/larger particles, and so on (i mean you know how physics is trying for a 'unified theory', afterall).

 

[Perfection]

Current electromagnetic theory is so wildly successful at describing the behavior of atoms and molecules to incredible precision, that it's implausible that there hidden phenomena that make electromagnetic theory invalid in describing things at the biochemical scale. That may sound like a bold statement, and indeed it is, but it is supported by the incredible progress and success of physics today. Whatever effect unknown subatomic processes have here simply are on the wrong energy scale to matter, or is just another way of looking at what we already know with current theory.

 

[El_Machinae]

The important thing to remember is the scale at which sentience operates.

As far as we can tell, sentience was built up by Natural Selection from ancestral sensory/reflex systems. So, it's built from the same processes that allow life, using the tools that Natural Selection worked with to build 'lower' systems.

Natural Selection seems to be very focused on working at the molecular scale, but not so much at the atomic scale. Biology is 'fuzzy' enough that isotopes don't really seem to matter. While the presence of CHNOPS is important, their specific isotopic concentrations don't seem to matter.

Sentience is built up, via selection, using the tools available to it. So, it seems very likely that the forces (even if they're EM fields) involved are ones that can be controlled by controlling the molecular inputs into the equation.