Philosophically, what are ways in which one can define "intelligence" in general?

Question

I've always been wondering, what with I.Q. tests and all, how one can better sense intelligence. Not intelligence as in how much we are measured to know, but how much of what we know is actually known in non-relation to informal testing and human comparison.

Essentially, how much of what we know makes us "smart" without comparing ourselves to peers and such, or people in other fields; think as in measuring our "smarts" as in applicability of potential or learning ability progression, brain limitations but desire, and application rather than simply "I know more". Simply, what if I could learn more but I had more trouble in the learning process itself? That would not mean I know less; it would mean I have more trouble learning or grasping something, but doesn't irrefutably prove limitations.

People define intelligence many ways:

1.A teenage girl can think a "hot" teenage boy is intelligent if he gives a goofy answer to a less goofy question, e.g., "Oh God, you are so smart! I never knew computers added by two's complement!"; this is a relation to another person;

2.A person who has never heard of science may be baffled by hearing things like, "Water turns to steam", "Things can exist for some time (virtual particles)", or even, "Your mind can fool you; you may see it but no one else around you can."; this is a relation to one who has no input or involvement.

Essentially, again, fairness can't be just in reasoning that a person from a farm with no technology is "doomed to succeed" because they are behind in socially-accepted knowledge forms.

"Smart" is a measurement in relation to another immediate factor (e.g. a person). How can we measure intelligent without that? Determine how smart I am with non-conformal I.Q.s, or socially-approved knowledge graphs? Theoretically I could be "smarter" than every other human on this planet, but I just don't have the immediate knowledge to prove this. And since "intelligence" is not 100% known, I may be smarter and just less capable; or I may be able to apply myself, but lack mental execution to succeed, leaving the question open for others to debate on whether, "I can but haven't yet", or whether, "I could, but my mind is limiting me."

Answer 1

Intelligence, much like other traits of a system, is relative. The statement "System A is intelligent" doesn't make any sense until you also specify compared to which system System A is intelligent. When we hear such statements we implicitly compare the system with other system that we are familiar with.

Let's say you have 2 systems - A and B. Both of these systems are able to perform a task X using certain amount of resources (lets say energy). Now if system A uses less energy as compared to system B, we can say that the system A is more intelligent than system B. But if both of the system performs different tasks then there is no way to compare their intelligence.

How can system A is able to perform the task using less energy then system B? The answer lies in the fact that system A is "exploiting" some patterns in the task to achieve it much more efficiently. Patterns acts as shortcut to achieve a target. The more patterns a system is able to exploit the more intelligent it will be.

Where does the pattern comes from? This is where the concept of learning comes into play. A system capable of learning will evolve its sets of patterns over a period of time to achieve its goals more efficiently.

In summary, you can say that being able to recognize patterns and exploiting those patterns to achieve some goals is what makes a system intelligent.

UPDATE:

I would like to add more clarification regarding what exactly an intelligent system has relation with patterns.

Any intelligent system must have the concept of observation. We humans observe the world around us using senses. A computer program observe the world by receiving input from user or other devices. Based on the observations the system can perform below activities (depends on the abilities of the system):

• System figures out which pattern is applicable based on the observation. The system will have a repository of patterns and when an observation occurs the system will need to figure out which particular pattern the observation is all about. A software system has fixed set of patterns and based on the input to the program there is a usually a if/else code which decides what pattern to use for the input. A human when sensing the environment decides which pattern the sense data belongs to - Your eyes see a car and brain decide that car is the existing pattern the visual data belongs to.

• System can apply (exploit) a pattern. Once the system has decided which pattern or set of patterns are applicable the system can apply the set of patterns to decide further set of actions and predictions. Once a software has decided that you have asked it to delete a file, it will perform the required action. Once a human have seen a car it will apply the "car" pattern to predict many things and may be perform some action.

• System learns new patterns / update existing patterns. The system has the capability to learn new patterns i.e based on the observation it can figure out new patterns in the observations and grow its repository of patterns and may even update its existing patterns due to new observations. The database of a software is the repository of patterns for the software. When you insert new data, update existing data or delete some data - the software just learned new patterns and updated existing patterns, although its pretty stupid learning - much like believing in whatever someone says. Machine learning is much more advanced form of learning than simple database operations.

• System can design new patterns from existing patterns. Welcome to the world of creativity. For this capability the system must have all the above capabilities. Look around the world and you will find there are so many things that we humans has created that never existed in the nature before. This is only possible because of this ability to create new patterns from existing patterns. Whole human civilization is just a complex set of patterns that we have created over centuries. Most of the time out brain is always busy in performing the above activities and I guess that this capability require less observations because the more the observations the more the brain will be busy with above activities. Probably that's why creative people try to find a place where their senses are not overloaded with observations.

Answer 2

This is a very good question, and not easy to answer. What intelligence purports to mean, seems to depend on what we're using the term for. The best purchase we can get on it without getting into a very long philosophical discussion is that "intelligence" as it's normally used refers to a broad collection of skills, some of which can be replicated by computers, some of which deals with vaguer concepts like common sense.

Artificial Intelligence provides a good background for the use of the word. In this case it's taken to mean an ability to solve problems in logic, and much of the challenge in producing artificial forms of intelligence deals with constructing a method for automated handling of problems which aren't necessarily amenable to calculation. It seems clear that the feats of memory which computers can perform with ease, aren't what most people think of when they talk about intelligence. This is arguable, but let's take basic read-from-disk operations as our starting example; few people would argue that following a mechanically-built rule to retrieve the contents of a box at a given address constitutes anything like "intelligence" in the usual sense. What we intuit intelligence to mean gets muddier the more complex this process of recalling information is, but this should make the case that even if you want to reduce intelligence to "a set of many operations of varying complexity that perform memory retrieval at appropriate times", then intelligence will still refer to something like that whole set of rules of retrieval and information available. Yet how we can go on to distinguish "less intelligent" from "more intelligent still isn't altogether clear; efficiency of our rule set may help to set lower bounds, but the ability to solve general problems isn't yet possible by rules alone, and some people would argue, may never be.

This reveals, I take it, that General Intelligence will normally be the target most philosophers will mean with the word 'intelligence': a general problem-solving ability that computation currently cannot deal with. Here we can go back to Gilbert Ryle (quoting from the Routledge Encyclopedia here- access is limited):

Ryle appears to have felt that he had to rebut the following style of argument, lest it be taken to give support to his belief in the Cartesian `ghost':

 1. There is intelligent behaviour.
 2. Intelligent behaviour is behaviour that is directed by intelligent thought. 
 3. Such thought would have to take place in a ghostly private medium. 

So:

 Talk about intelligence is talk about what goes on in the ghostly medium.

Ryle's strategy in both books was to reject (2), denying that intelligent behaviour must be controlled by a process of considering propositions. When we call behaviour intelligent we are speaking of the manner of the behaviour itself, most importantly of its responsiveness to changes in the conditions under which it occurs.

Now many philosophers (as the REP later points out) would reject that (3) is a reasonable component for this argument, but whether consciousness is a necessary or sufficient component of intelligence goes far afield of the intent of your question. Instead, I think we can take Ryle's aspect of responsiveness here to be what we're after, if we want to continue the project of ranking one intelligence as "better" or "worse" than the next.

This may not be wholly satisfactory- it does seem to move the question a little bit away towards what responsiveness means, but given the above look at artificial intelligence, the contrast suggests that not just any stimulus-response will do. Given the context of your question, the combined ability to recognize and identify complex stimuli and provide the most efficient response in pursuit of some goal is the first signal of intelligence; if we want to know whether this is a case of general intelligence, I can only imagine we'd want to see novelty in the efficiency of the response as well.

Answer 3

i like the definition that intelligence is what intelligence tests try to measure.

people can and do use the term differently to that, but the above is how the term is used in psychometrics, or "psychological measurement". two objections may be:

1. my intelligence is immeasurable
2. or, psychology is not a good enough science to redefine our terms.

but intelligence is for me a behaviour and so [one would think] measurable. and i would only suggest that the above definition is our best definition of intelligence, not the only one.