|
Author
|
Topic: The Characterization of Intelligent Causation
|
aiguy
Member
Member # 3736
|
posted 31. March 2007 17:50
Daniel,
quote:
The question then, is: Do the earth's hydraulic, gravitational and weather systems show evidence of intelligent design?
No, I think the question is "Are the earth's systems intelligent" per the characterization of intelligence you have offered.
quote:
Attributing "intelligence" to the water and clouds is just as irrelevant as attributing "intelligence" to DNA, any other biological system or any human invention for that matter. We are talking about whether or not these systems have intelligence as their first cause, not whether they are intelligent in themselves.
If you claim that something has "intelligence" as a cause (proximate or ultimate), you need to be able to show that whatever enables the mental abilities of human beings is the same thing as whatever caused the phenomenon in question. Without a specific model of both of these causes - at some level of abstraction that is more specific than simply "solving problems" or "processing information" - it seems impossible to support this identity relation. All sorts of things can be said to process information, and all sorts of things can be said to solve problems, but we don't normally think of all these things as intelligent, and we don't consider all of them to share the same cause.
[ 31. March 2007, 17:51: Message edited by: aiguy ]
IP: Logged
|
|
Melvin H. Fox
Member
Member # 1684
|
posted 31. March 2007 18:39
I rather like Daniel’s idea (intelligent agent = problem solver). As for the problem solving river; does it have a self-determined, self-established value system? Does the river recognize a problem (a situation requiring a solution)? Does the river have any ability to solve (find answer) the problem? We should assume it does not unless we can demonstrate that it does. Once again I agree with Daniel, the river is at best a solution.
The value system is of the utmost importance. Problem solving is all about value judgments. If the agent can’t judge the rightness or wrongness of a situation it can’t begin to recognize the problem, let alone find the solution. The value system can’t be mandated by any other agent. This is so because any solution determined with the mandated system as its basis would belong to the agent issuing the command and not to the subservient (programmed) agent.
So then, three necessary conditions for the existence of intelligence, or the ability to problem solve, are free-will, cognition, and limited power.
If we could demonstrate that the river decided that going to the ocean was the right thing to do and that going down hill will lead to the ocean and that it has the energy to get there, then we might proclaim it to be a genius. Or, we might not. After all, our value system could differ from that of the river and by such a degree as to tag it an idiot.
-Mel
IP: Logged
|
|
aiguy
Member
Member # 3736
|
posted 31. March 2007 20:02
Mel,
quote:
As for the problem solving river; does it have a self-determined, self-established value system? Does the river recognize a problem (a situation requiring a solution)? Does the river have any ability to solve (find answer) the problem? We should assume it does not unless we can demonstrate that it does. Once again I agree with Daniel, the river is at best a solution.
How might one demonstrate that the cause of biological complexity was self-determined, or has a self-established value system, or "recognizes" a problem rather than simply solves it? As far as I can tell, it would be quite impossible - after all, philosophers have recognized for the past few millenia that the problem of other minds has no empirical solution, even when the candidate is standing right in front of you. Solving this problem for some hypothetical entity to which we have no access would seem even more certainly impossible.
quote:
So then, three necessary conditions for the existence of intelligence, or the ability to problem solve, are free-will, cognition, and limited power. If we could demonstrate that the river decided that going to the ocean was the right thing to do and that going down hill will lead to the ocean and that it has the energy to get there, then we might proclaim it to be a genius. Or, we might not. After all, our value system could differ from that of the river and by such a degree as to tag it an idiot.
An ant solves the problem of finding a pheremone trail back to the nest. Does this mean the ant has free will? Does it mean that the cause of its nest-finding behavior is somehow the same thing as the cause operating when a human being does it?
As for self-programmed versus being programmed by other agents, I believe this whole issue is terribly confused. ID claims that human beings are designed, yet are fully intelligent and volitional. Doesn't the fact that they have been programmed by their designer negate this requirement of being self-programmed? If you think science is capable of detecting that human beings act out of contra-causal free will, you should review the literature (cf Benjamin Libet's studies). We can demonstrate no such thing.
[ 31. March 2007, 20:02: Message edited by: aiguy ]
IP: Logged
|
|
LifeEngineer
Member
Member # 3446
|
posted 01. April 2007 09:47
Quote: As you've described it here, this "basic information processing model" appears pretty much tantamount to "Turing computable". The claim that all processes are Turing computable may or may not be true, but even if it is, it isn't much of an explanation of anything. It certainly doesn't help us make a meaningful statement about an identity relation between two processes.
Clearly the basic information processing model I have described is not original and the name Turing machine would be as good as any other to describe the internal processing of the information processing model. I don't recall exactly how Turing addresses the external components of information processing. The key point being, however, that this information processing model is a standard widely used type or class of model or paradigm.
I believe there is mathematical evidence that not all relationships are Turing computable and that is not the claim being made here. The claim here is that any intelligent goal directed behavior can be modeled, simulated and scientifically analyzed using an information processing model.
There are recognized limits to human knowledge and human scientific knowledge. The fact that there may be mathematical limits or limitations to what can be computed or simulated by Turing machines does not imply that Turing machines are not useful or appropriate for the scientific analysis, modeling, and simulationg of intelligent behavior.
You appear to have a rather unconventional view of scientific explanations. Formal scientific explanations of compex phenomena like intelligent behavior arise when something can be modeled and simulated using logically consistent sets of testable predictive theories.
Scientific explanations of intelligent goal directed behaviors are provided when the behaviors can be modeled and simulated using the best currently available testable predictive theories. If such theories are produced using information processing models or paradigms (involving Turning machines) then these models have produced scientific explanations. Furthermore, testable predictive theories produced using these models do not need to be perfect, they simply need to be better than models and theories produced using alternative models and methods.
It should be obvious that both intelligent human behaviors and intelligent biological change procedures can be and have been modeled and simulated using Turing based information processing models. The only three issues in dispute are 1)Can these Turing based models and simulations be associated with predictive theories?, 2) Do the Turing based models and simulations of evolution and human behavior use the same definition and concept of intelligence? and 3) Are there any existing successful alternative definitions or approaches?
It is important to keep in mind that essentially all successful computer models and simulations of both human behavior and biological evolution are and have been based on variations of the information processing model or paradigm. The only real debate/conflict arises from recognizing the predictive hard science theories associated with these models.
IP: Logged
|
|
Melvin H. Fox
Member
Member # 1684
|
posted 01. April 2007 10:20
Ants, as you point out, have a more sophisticated navigation system then do rivers. Does the use of the combination of integration vectors, visual landmarks, and pheromone trails, in and of itself, imply the individual ant is an intelligent agent? That is hard to discern in this case because it is not clear weather the situation is a problem or a task. Unintelligent agents (computers) can be trained to perform tasks. You see, if the path is familiar and clear, then there is no problem. If the philosophers of the past few millennia are correct, then the problem of knowing the ant’s mind has no empirical solution. Forgive me that I do not wish to roll over and play dead here. I suggest, here below, two initial experiments to investigate the individual ant’s ability to solve problems.
Experiment 1: Remove the ant from its frame of reference (all familiar visual landmarks and pheromone trails). However, construct no impassable obstructions between the ant’s location and its frame of reference. Can the ant find its frame of reference? Does each ant presented with this predicament follow the same exact procedure or do tactics very? If the individual ant can find its way home and tactics very from ant to ant, then we would have empirical evidence for intelligence. What is more, we could ascribe this intelligence to the individual ant. If instead, all the ants used the same algorithm and each ant found its way back to the frame of reference, then we would have empirical evidence for intelligence, BUT we could not necessarily ascribe it to the individual ant.
Experiment 2: Take the ant and place it in a cage (box with an open top and sides so smooth that it can’t clime up them). Provide the ant with food and water. Also, provide the ant with various building blocks (leaves and what not) that the ant has the power to manipulate. Here is a true problem for the ant. Does the ant desire to leave the box? Can the ant recognize the problem of the obstruction, the usefulness of the building blocks, and its own power to manipulate them in such a way as to overcome the obstacle? If the answer is yes to all of these questions, then the ant will leave the box and thus demonstrate its intelligence empirically.
"Doesn't the fact that they have been programmed by their designer negate this requirement of being self-programmed?"
That an agent does, by individual purposeful choice, program implies that that agent can solve problems which further implies that that agent is intelligent.
-Mel
IP: Logged
|
|
LifeEngineer
Member
Member # 3446
|
posted 01. April 2007 10:24
Quote: If you claim that something has "intelligence" as a cause (proximate or ultimate), you need to be able to show that whatever enables the mental abilities of human beings is the same thing as whatever caused the phenomenon in question. Without a specific model of both of these causes - at some level of abstraction that is more specific than simply "solving problems" or "processing information" -
This is not correct. This would only be correct if you were assuming that intelligence is a single physical force or phenomena. Science does not reqiure that 'intelligence' be so defined and in fact science has no method of proving that even things like gravity that we accept as a single force are in fact a single physical phenomenon or force.
Intelligence is a useful abstraction to be defined by sceintists developing predictive theories. If the same abstractions and definitions can be used for writing music, solving math problems, and evolving birds then, as scientists, we can assert that a common 'intelligence' was involved in all three.
Much if not most of the confusion and conflict regarding defining intelligent causation appears to arise from confusion regarding the scientific requirements for defining an abstract concept. It is impossible to define intelligence only when unrealistic and inappropriate requirements are imposed.
When I was analyzing this issue I found it convenient to use the term process to describe a phenomenon or abstraction believed to involve a single physical phenomenon like gravity and the term paradigm to describe an abstraction that can take multiple physical forms. Since bridges and airplanes can be constructed from a variety of different materials the abstractions involved are paradigms. Intelligence can involve a wide variety of different physical forms, therefore it is defined by science as paradigm. There is probably better terminology that could be used, but the concept should be clear.
IP: Logged
|
|
aiguy
Member
Member # 3736
|
posted 01. April 2007 14:31
Mel,
quote:
Ants, as you point out, have a more sophisticated navigation system then do rivers.
Maybe, but rivers manage to find an optimal path to sea, considering various different variables. While at each point they look for local minima of altitude, they also avoid boulders and harder soil in favor of eroding softer soil...
quote:
If the philosophers of the past few millennia are correct, then the problem of knowing the ant’s mind has no empirical solution. Forgive me that I do not wish to roll over and play dead here.
Nobody is asking for that, I assure you! My point is simply that we can't do experiments on the "agent" assumed by ID to be responsible for creating life. In light of this, my question is how can we characterize intelligence such that we can empirically determine that this cause, and the cause of human behavior, is the same thing.
Your ant experiments provide one particular operationalized definition for intelligence. (I actually believe ants would fail your experiment #1, and so might people, but that is really beside that point). We can't apply this particular definition to the cause of life, so I don't understand the utility of this exercise in the context of this discussion.
quote:
That an agent does, by individual purposeful choice, program implies that that agent can solve problems which further implies that that agent is intelligent.
But computers can write programs too. Do they use "individual purposeful choice" to do so? I will answer this question "yes"; perhaps you disagree, but you have offered no means by which we could decide who is correct. The fact that computers are programmed by something else ought not exclude them from being called intelligent, any more than the fact that human beings are programmed (by teachers, peers, and perhaps - if ID is correct - by an Intelligent Designer) ought to exclude them from being called intelligent.
IP: Logged
|
|
aiguy
Member
Member # 3736
|
posted 01. April 2007 14:34
quote:
Clearly the basic information processing model I have described is not original and the name Turing machine would be as good as any other to describe the internal processing of the information processing model. I don't recall exactly how Turing addresses the external components of information processing. The key point being, however, that this information processing model is a standard widely used type or class of model or paradigm.
A Turing machine is a mathematical description of digital information processing itself, not a model or paradigm that is "used" in computer science. There is no such thing as a computer system that is not a Turing machine.
quote:
This is not correct. This would only be correct if you were assuming that intelligence is a single physical force or phenomena. Science does not reqiure that 'intelligence' be so defined and in fact science has no method of proving that even things like gravity that we accept as a single force are in fact a single physical phenomenon or force.
Nothing I said had to do with identifying intelligence as "a single force". I explicitly said that a characterization at some level of abstraction is what is required. Saying that two processes are both Turing computable does not imply that those two processes are the same thing in any meaningful sense at all.
IP: Logged
|
|
Daniel Smith
Member
Member # 3004
|
posted 01. April 2007 16:07
aiguy:
quote:
How might one demonstrate that the cause of biological complexity was self-determined, or has a self-established value system, or "recognizes" a problem rather than simply solves it? As far as I can tell, it would be quite impossible - after all, philosophers have recognized for the past few millenia that the problem of other minds has no empirical solution, even when the candidate is standing right in front of you. Solving this problem for some hypothetical entity to which we have no access would seem even more certainly impossible.
You are right, we can't directly study what goes on in someone else's mind.
To directly observe "intelligence" is therefore pretty much impossible.
We can however look at it's results (words, actions, designs, etc.) and easily see that this unobservable phenomena called "intelligence" was definitely involved (see Paley's watch). For example, a "machine" of any type is always assumed to be the direct result of intelligent planning.
So what exactly are the properties which allow us to infer that unobservable intelligence was involved in the origin of observable objects?
I'd say that they (the objects) show evidence of being the result of planning, purpose and creative problem solving.
Let's continue the discussion about the river:
The water molecule is a marvelous thing; in liquid form it is heavy enough to take the path of least resistance to the lowest point it can. When it is in vapor form it is light enough to be carried on the winds, and when it is in solid form it resists both and essentially becomes it own self contained storage device.
So does the design of the water molecule show evidence of planning, purpose and creative problem solving?
It looks that way to me. Couple that together with the fact that it is essential for life and part of a much larger cycle which works continually, exactly as it must, in order to sustain said life and I think the argument becomes overwhelming.
Now as to the other question: Can intelligence create intelligence? All I'll say is that we're trying. Surely those involved in artificial intelligence (I assume from your moniker that you are) must know what it is they are trying to replicate?
What are the hallmarks of intelligence in the AI world?
Are planning, purpose and creative problem solving involved?
IOW, how will they know when they have succeeded?
IP: Logged
|
|
aiguy
Member
Member # 3736
|
posted 01. April 2007 17:10
Daniel,
quote:
You are right, we can't directly study what goes on in someone else's mind. To directly observe "intelligence" is therefore pretty much impossible.
We can't observe others' phenomenal consciousness, certainly, nor can we observe free will. Whether or not we can observe "intelligence" depends entirely on how it might be defined, which is the question at hand.
quote:
We can however look at it's results (words, actions, designs, etc.) and easily see that this unobservable phenomena called "intelligence" was definitely involved (see Paley's watch). For example, a "machine" of any type is always assumed to be the direct result of intelligent planning.
This completely begs the question, which is, how is it that we are characterizing this thing, "intelligence"?
quote:
So what exactly are the properties which allow us to infer that unobservable intelligence was involved in the origin of observable objects?
We can hardly infer something until we say what it is we are trying to infer. Are you sticking with volition and intentionality? What is the method you propose for establishing when these traits are present in human beings, and then in the cause of biological complexity?
quote:
I'd say that they (the objects) show evidence of being the result of planning, purpose and creative problem solving. So does the design of the water molecule show evidence of planning, purpose and creative problem solving?
It looks that way to me. Couple that together with the fact that it is essential for life and part of a much larger cycle which works continually, exactly as it must, in order to sustain said life and I think the argument becomes overwhelming.
What sorts of evidence might show this? How do we know when something as been planned purposefully and creatively, as opposed to, say, planned uncreatively and with no purpose in mind?
If we're simply talking about how things strike us subjectively, then we should have a different discussion - one that has nothing to do with scientifically supported claims. Saying that water molecules look designed to you is really not a sufficient criterion: They don't look designed to me - how do you propose we resolve our disagreement?
quote:
Now as to the other question: Can intelligence create intelligence?
I didn't raise that particular question, and I don't see the direct relevance to the topic here. Again, until we decide on a particular characterization of intelligence that can be applied in both the context of human behavior and in the context of the creation of life, we will not be able to assess any of these questions. The characterization must enable us to objectively and unambiguously decide how to categorize any arbitrary system as to whether or not it possesses the property of intelligence, and demonstrate that the same characterization applies to humans and to the cause of biological complexity.
quote:
All I'll say is that we're trying. Surely those involved in artificial intelligence (I assume from your moniker that you are) must know what it is they are trying to replicate? What are the hallmarks of intelligence in the AI world?
There are no particular "hallmarks"; AI attempts to replicate human competence in all sorts of different tasks. At the inception of AI, John McCarthy provided the definition of "intelligence" that has served ever since: An intelligent behavior is one which we would tend to call intelligent if a human being did it. In other words, McCarthy pointed out that it makes no difference at all if we call something intelligent - it is a loose descriptive label that adds no information beyond a description of the ability itself.
quote:
Are planning, purpose and creative problem solving involved?
Planning is certainly one type of skill that AI seeks to replicate, and AI systems are very good at planning. "Purpose" is something we project onto a system, not something objectively detectable (the purpose of the river is to move water to the sea; the purpose of a rock is to sit on the ground). Likewise, whether or not "creative problem solving" is occurring is not a matter of fact, but only of definition.
quote:
IOW, how will they know when they have succeeded?
An AI system is successful if it does a particular task successfully. There is no special criterion for achieving "real" intelligence; each system simply does what it does.
IP: Logged
|
|
LifeEngineer
Member
Member # 3446
|
posted 01. April 2007 19:38
Quote: Nothing I said had to do with identifying intelligence as "a single force". I explicitly said that a characterization at some level of abstraction is what is required. Saying that two processes are both Turing computable does not imply that those two processes are the same thing in any meaningful sense at all.
Your comments are rather confusing and offtrack from the discussion. It appears that you are agreeing that an abstraction defined in the context of the defined information processing model or paradigm could be applicable to both human behavior and evolutionary biology. You appear to be attempting to create the impression that you are disagreeing with what has been presented without in fact presenting any contrary argument.
So there is no confusion, the information processing in all computer systems can be viewed or described as or in terms of Turing machines. Also all computer systems or essentially all computers systems can be characterized as purposeful or teleological or goal directed. All computer systems can be described as compatible witht he information processing model or paradigm I defined.
Given this framework, we can now offer definitions of the intelligent causation abstraction that can be (and have been) used for human behavior, biological evolution and computer behavior.
IP: Logged
|
|
Melvin H. Fox
Member
Member # 1684
|
posted 01. April 2007 20:13
I never met a computer that did not do exactly what it was told to do and only what it was told to do. If we ever create one that does not, then we will very quickly regret having done so.
I agree that many humans would fail both of my proposed experiments. Why I know some who can’t find their way back from the salad bar. This does not mean they are unintelligent, only that they have failed to give us any empirical evidence of it in that situation.
I do not wish to derail the discussion. You do not want to accept the intelligence equals problem solving ability definition because it is not useful for this line of thought. I accept and defer to your better judgment on this matter. Carry on.
-Mel
IP: Logged
|
|
aiguy
Member
Member # 3736
|
posted 01. April 2007 20:23
quote:
All computer systems can be described as compatible witht he information processing model or paradigm I defined. Given this framework, we can now offer definitions of the intelligent causation abstraction that can be (and have been) used for human behavior, biological evolution and computer behavior.
You seem to be saying that we should characterize "intelligence" as "that which is Turing-computable". In other words, in answer to the question "How does a human being design a watch?", the answer sholud be "By means of some process that is Turing computable". Given that everything is Turing computable as far as we know, you are saying that everything is intelligent.
IP: Logged
|
|
aiguy
Member
Member # 3736
|
posted 01. April 2007 23:54
Mel,
quote:
I never met a computer that did not do exactly what it was told to do and only what it was told to do. If we ever create one that does not, then we will very quickly regret having done so.
Uh-oh - too late! Many computer systems do more than they are told to do, of course. There are systems that learn from experience, systems that incorporate random input to search for novel solutions to problems, systems that guess what they ought to do based on incomplete information, and so on.
quote:
I do not wish to derail the discussion. You do not want to accept the intelligence equals problem solving ability definition because it is not useful for this line of thought. I accept and defer to your better judgment on this matter. Carry on.
I've been thinking about minds and brains since I was a little boy (a very long time ago). When I heard about Intelligent Design theory, I imagined that people would be talking about what a mind is, and what intelligence is - since, after all, ID offers "mind" as an explanation of things we see in the world. To my great surprise, I find that ID literature ignores this entirely. I think once one gives some serious thought to the problems of volition, representation, sentience, and so on, the claims of ID appear at once to be facile and unscientific.
IP: Logged
|
|
Melvin H. Fox
Member
Member # 1684
|
posted 02. April 2007 07:19
quote:
Uh-oh - too late! Many computer systems do more than they are told to do, of course.
Bhaa!
quote:
There are systems that learn from experience,
These systems gather information as they have been told to do and store it where and how they have been told to do.
quote:
systems that incorporate random input to search for novel solutions to problems,
Novel solutions, sure, but the search algorithms are prescribed and any modifications made by the computer to these algorithms are done in a prescribed way.
quote:
systems that guess what they ought to do based on incomplete information, and so on.
Guess? This is an interesting choice of word. The computer has no inkling of what a guess is. It is told exactly how to select what to do. This selection may be as a result of some “random” process as set down in the programming but the computer can’t differentiate those instructions from the type of instructions we would classify as a non-guess.
Look, call me when you tell the computer to guess because you have programmed it to guess and its response is to print to screen; “No! Guessing is wrong.”
-Mel
IP: Logged
|
|
|
Printer-friendly view of this topic