Is Intelligence Relative?

Taylor makes an interesting point in reaction to my post about The Turing Test not working. This got me thinking. Is there a universal standard of intelligence, or is intelligence relative?

What is "intelligence" anyway and how can it be measured? Is intelligence simply the ability to solve logic problems, or math problems? Or is it the ability to think creatively? Or to be irrational as well as rational? What exactly is intelligence anyway? Isn't intelligence more of a continuum than a set of discrete jumps, and if discrete, then how do we define the boundary between one quanta and then next? The concept of measuring intelligence is always inherently biased -- for example do the SAT exams, or IQ tests, measure "intelligence" or simply the ability of a certain audience to solve a particular set of problems? Would Picasso be judged as "intelligent" on the SAT? Similarly, would a painting-test result in Einsteing being judged as intelligent? Intelligence, like beauty, is in the eye of the beholder. So ultimatley any standard for intelligence would be totally relative to the party inventing the standard. We might be able to devise a generic standard of human intelligence, but to do that we would probably have to test for many things, not just logic. This would take the individual human subjectivity out of the Turing Test -- there would be no individual human judge per se -- but would substitute that with the "collective subjectivity" of the group that devises the standard of measurement -- so isn't this just replacing one weakness with another? In the end, wouldn't this approach just result in the Turing Test effectively testing the intelligence of the group that devises the standard for measuring intelligence? If they come up with a really dumb standard almost every machine will seem intelligent, if they are very very smart and come up with a really subtle test then perhaps only some small percentage of machines will be judged as intelligent. But in the end, it still comes down to the beholder not the machine being tested. Intelligence is completely relative -- If I am super-brilliant then Joe Schmoe may seem really dumb. If I am a complete moron, then Joe Schmoe may seem really intelligent. Is there a universal standard for intelligence? How can we define it in a non-anthropocentric, non-culturally-centric, non-brain-hemisphere-centric way?

Comments

[Bill Brown]

Intelligence is general to humans. The determination of "intelligent" is what is binary.

Intelligence (and especially reason) only apply to humans. Reason is not something that computers or other animals possess. Computers are dumb: processing instructions without criticism or independence. Any time they seem rational, it's really a codified chain of reasoning that they are following. The gap between processing instructions and reasoning is insurmountable. The best that we can hope for is extremely complex codified chains of reasoning.

The terms "rational" and "irrational" only apply to entities capable of exercising reason—and conversely capable of choosing not to exercise reason. Therein lies the heart of the matter: choice. Reason is nothing without volition. Computers don't (and, I think, can't) possess volition. Some animals have a very limited faculty of choice, but most act from hardwired instructions. Those animals that do have volition don't have reason—i.e., volition is not sufficient for reason, but is necessary.

If I could objectively measure intelligence, I sure wouldn't be sharing it here. I would have written the research paper to end all research papers. The best I can do is to say whether someone is more intelligent than another.

As for the major question you ask regarding Turing tests, I think its a red herring, a blind alley. Computers cannot have intelligence by their nature. It's possible that organic computers in the future might be able to learn from their programming, but I think it's highly unlikely given that significant gap between sentience and intelligence. Whether organic computers could ever achieve intelligence is open for speculation, but I know that computers as we know them never will.

[Nova]

How can we determine whether "intelligence" is on or off? If all we define intelligence by is a binary decision, then really almost anything could be considered to be intelligent -- a thermostat for example! What is the standard for "intelligence?" Once again, I think we end up defining intelligence as "humanlike intelligence" which is quite unique from "computer intelligence," "parrot intelligence," "chimp intelligence," "sea slug intelligence," etc. Why? What makes "humanlike intelligence" so "humanlike?" I think that unlike other forms of intelligence we humans have a special ability to engage in both rational and irrational thought. Computers can only be rational. Many animals and insects cannot really be rational. But humans can be both. This is special. We can simulate any Turing machine with our brains (given enough time and paper!), or we can play improvizational jazz, write poetry, or invent imaginary worlds. We are cool and logical, we are also intuitive and unpredictable. The question is: can this be measured? Can we detect it? What kind of test could we devise for this that would enable us to objectively determine whether something had "humanlike" intelligence? One half of the test would have to be logic problems, another half would have to test creativity and "non-logical" intelligence. How would that half work?

[Bill Brown]

Anthropocentric means focused or specific to humans. Non-anthropocentric would be the opposite, thus inclusive of all creatures.

I doubt that computers or AI could ever match, much less surpass human intelligence. It's just too multifaceted and too diverse. In certain of these facets, computers can be programmed to do much better than humans (not a controversial opinion) but intelligence is the whole, the sum. We don't understand intelligence enough to replicate it programmatically and I'm pretty sure that our understanding is still fairly far off.

It's universal in that intelligence is not cultural or specific to any single society—it's an essential part of humanity. However, we can say that a particular human is more or less intelligent than another and we can say that one person is intelligent and another is not. I don't know that I can explain it any better or differently than this and my last comment.

[Taylor House]

When was this ever about animals? I'm concerned with the intellgence of HUMANS when compared with COMPUTERS and AI, which, quite possibly, could one day surpass our own intelligence. Remember, so far it's only theory.

(Intelligence is universal but an all-or-nothing generalization? Explain.)

[Bill Brown]

How can we define it in a non-anthropocentric, non-culturally-centric, non-brain-hemisphere-centric way?

I think we can get rid of the non-anthropocentric part of that. The intelligence of humans is beyond any other animal and devising a test to measure all creatures' intelligence equally is both futile and a disservice to humans. Our capacities are completely unique in the animal kingdom.

That said, is there a universal measure of intelligence? No. Is intelligence universal? Yes. Intelligence is so imperfectly understood at this stage of the game that we can't even accurately measure our own, much less generalize it across cultures and societies. That doesn't mean, of course, that it doesn't exist—just that we currently cannot measure it.

I'm inclined towards Howard Gardner's view of multiple intelligences, though I prefer to think of it as multifaceted intelligence since his separation ignores (or minimizes) the considerable fuzziness between the divisions.

I'm also inclined to think of intelligence as an all or nothing generalization; there is definitely a difference between those who are and those who aren't. Within each division, though, there are gradations but we can't define it more precisely than cardinally: he is more intelligent than him but less than her.

[Taylor House]

INTELLIGENCE is the ability to excel, whether it be art (Picasso) or physics (Einstein). Being INTELLIGENT is the effect to which you use your ability for INTELLIGENCE.

Let's say that this basic set of functions every human is born with isn't set at zero when you're born, but varies: one child may be born intelligent (Level 60) and another is born less intelligent (Level 10). (These could be caused by genes, mutations, etc.) Level 60 chooses not to expand his mind, and remains a Level 60, but Level 10 chooses to read, think and make himself more aware, eventually surpassing the Level 60.

I believe everyone (from level 1 to 100) is born with intelligence - the ablity to become intelligent. Some realize it's power and use it to their advantage, and become more intelligent.

(Picasso and Einstein may have been on the same "intelligent level," but choose to express their genius in different ways. I don't think there is any way to compare or measure these men in their respective fields.)

[Josh Kirschenbaum]

There is also the difference between Intelligence and Intelligent. I believe that Intelligence is not a fuzzy state- you either have Intelligence or you don't. I have Intelligence - my chair does not.

Intelligent is a relative term. I know that I am more intelligent than some people, and I know that there are people who are more intelligent than me. This has nothing to with Intelligence- we all have that.

The Turing Test was originally designed to detect Intelligence- some of the questions were purposefully "simple"- which is why some humans have failed the test...they miss the simplicity, expecting the test to be about Intelligent, not Intelligence.

So- Intelligenct IS relative, and I don't think it can ever be "standardized" beyond a system that classifies intelligence for various functions and tasks.

But Intelligence is absolute. It is on or off.

j.