[nudephil (OP)]

A question here from Tristan:

I was listening to an episode where you were talking about intelligence and IQ and was wondering what you thought of recent attempts to give a more rigorous definition of intelligence? For example: https://definitionmining.com/index.php/2018/01/03/intelligence/

What do you think?

[alancalverd]

"The ability to compare concepts" is, I think, too vague to count as a definition, and still doesn't give you any clue as to how you might measure it.

I'm beginning to think that it's a nonmeasurable quality like beauty, that we ascribe to animals that have demonstrated an ability to learn and perform tasks with a significant mental content. This means that it is not arbitrarily  comparable between individuals - you have to define the task.

IQ tests only appear to be free from social context if you are actually immersed in that context. Consider the narrative "John and Jane set out to...." If you were brought up as a Taliban, you would know instinctively that Jane couldn't or shouldn't. Some years back there was a lot of interest in a South American tribe that had no concept of number greater than 3, and whose only directional vectors were "up river", "down river", "towards river" or "away from river", so they would have failed any conventional IQ test but had thrived in the most uncompromising jungle for thousands of years.   

My father had a friend who, though desperate to fight for King and Country,  couldn't march. The ability to coordinate right arm with left leg was completely beyond him. Within days of being written off as mentally unfit for combat, somebody noticed the letters "PhD", repeated, on his personnel record. He was whisked off to a secret location where his doctorates in maths, physics and German were put to better use. There's more than one way to win a war, and more than one form of intelligence.   

[David Cooper]

IQ is a shallow test of intelligence which is substantially just a race against the clock. People with very high IQs exhibit a wide range of bonkers political beliefs, so it's clearly failing to tell the full story. That said though, if you compare two groups of people with significantly different IQs, you will find greater problem-solving skills in the higher IQ group, but there may be some people in the lower IQ group who outperform some members of the high IQ group on hard tasks. An imperfect intelligence test could be right on average, but wildly wrong with some individuals.

I've been studying intelligence for a long time and can tell you exactly what it is. It's primarily about modelling and simulation. The more detailed the models in your head and the more rigorously you run them as simulations, the more accurately you can predict how things will interact and the consequences of different actions. A dog with a long stick in its mouth is bad at running simulations, so it just tries to get though a doorway repeatedly, typically making the same mistake every time until it drops the stick, picks it up nearer one end by luck, and then the only side that catches helps to angle the stick such that it finally goes through the gap. Very young children can be just as slow to solve problems of that kind, but they have the capacity to improve their performance far beyond other species. Simulation is much more central to intelligence than applying reasoning rules.

I've discovered something else though, and that is that most people apply a bad algorithm which they start out with in childhood and never switch away from. They simply collect beliefs from trusted sources and never question them. This is a remarkably successful strategy because it eliminates most of the need to think, taking people to high academic levels just by dint of them being able to parrot what they've learned and making sure they conform to the required beliefs of the establishment they work in. That is how it works in almost every field, and because most of the established beliefs in most academic fields are correct, or at least reasonably accurate, most things run smoothly. There are situations though where standard beliefs are plain wrong and where they have been shown to be so by rigorous mathematical proofs, but the proofs are simply rejected by the establishment on the basis of authority: they must be wrong because they go against what all the qualified experts say, while the qualified experts are experts precisely because they've taken on the required beliefs, while those who rejected them were not awarded qualifications. That's where the strategy breaks and science fails to self-correct.

A solution to this will soon be available though: the era of AGI is about to begin. Machines that simply apply the rules of mathematics rigorously and systematically to all things without error will start to test everything that scientists, politicians, philosophers and mathematicians have been up to, finding out which beliefs actually stand up to scrutiny and which are horribly broken. AGI will do intelligence properly, modelling and simulating things without breaking rules on an ad hoc basis to accommodate the broken theories of celebrated icons of physics, but by testing all ideas with the exact same rules. We have some humans who can do this already in restricted domains but who are simply shouted down by the establishment whenever they find faults, but the establishment will not be able to withstand the onslaught that's going to come their way once everyone has AGI on their desk and in their pocket, tearing their faulty beliefs to shreds at every turn. The fun is about to begin: the build of the world's first prototype AGI system was completed in November and it's nearly out of a long testing and debugging phase. This kind of software will be able to score individuals for their actual intelligence, so we will all find out exactly where we stand.

[alancalverd]

Just remembered a very good definition of intelligence: the ability to surprise the examiner. Difficult to quantify, but obvious when you see it.

And another: constructive laziness, likewise.

For a supreme example of both, see the South American tribe I mentioned earlier. And for a counterexample, ask any sudoku player why he does it.

[Bored chemist]

As far as I can tell, intelligence is the ability to solve problems.

[duffyd]

The emphasis on measuring fluid intelligence is the way to go.

[syhprum]

" a nonmeasurable quality like beauty,"
There is a unit to define beauty the Milihelen the amount of beauty launch one ship

[alancalverd]

Millihelen, surely? I think you may be harking back to the Sixties when the unit was first published, along with the concept of an attoskirt being the quantum of decency. It was all very confusing, due to the constant haze of marijuana, but without a doubt the best ever time to be alive.

[duffyd]

" a nonmeasurable quality like beauty,"
There is a unit to define beauty the Milihelen the amount of beauty launch one ship

Tremendous progress is being made. Just distinguishing between fluid and crystallized intelligence is a major breakthrough. Unfortunately, testers may not be uniquely qualified to perform them properly, they are so complex.

[Edwina Lee]

Intelligence is a word which refers to the general abilities of thinking. It is an open word which requires no precise definition, and a precise definition would undermine the openess of the word.
IQ (Intelligence Quotient) is a term originated from a test devised to quantify intelligence. The work was considered discredited after decades have passed.

IQ tests are good at predicting university sucesses. One might as well say that IQ tests are a type of aptitude tests for predicting performance for particular settings.