|
|
Post by Baron von Lotsov on Jan 9, 2024 1:28:58 GMT
Like two economists in a room, you often find two psychologists profoundly disagree with one another. This is what I observed regarding intelligence tests. One says yes you can and it is easy peasy to do, and the other says it is bullshit.
The one who says it is easy, says you just randomise the questions and then after running many trials you can compute a scale which is sensible. The other says no you can't because it depends on so many things, like people's intelligence is affected when they are tired, some by the time of day and in any case what you are measuring is the speed at which people can answer questions, and speed is a poor proxy for intelligence.
I suppose we could dig a little deeper and ask how do you define it anyway. Also is it really that intelligent to measure someone's mental abilities by a single dimension? In academia a similar thing is practised where an academic's ability is measured by something called an H Index. This this H Index is rather arbitrary and is prone to people cheating the system as well. I can see the pull factor in inventing these indexes. If someone goes for a job interview the guy selecting the candidates just needs to select the highest score, and job done. The downside of course is if the system of measurement is flawed the that's like a single point of failure for the entire economy. You could make the same argument for national qualifications too.
|
|
|
|
Post by Orac on Jan 9, 2024 11:39:47 GMT
I feel there is a lot of unnecessary faffing and fluffing about the exact definition of intelligence.
IMHO Intelligence is the psychological ability to provide the one correct answer when an incorrect answer is far more likely. Accurate recall and retention is part of this ability, but the bit that seems to be most valuable is the part that creates an answer within a fixed (for all) information domain. This ability, if it is measured correctly, should be strongly linked to real world benefits for the holder. The acid test for IQ tests is that those with high measured IQs should in show more aggregate success in solving real life problems for themselves (ie have more success in getting what they want). This seems to be the case. Emotional intelligence almost certainly exists but the ability is a little tricky to capture objectively.
|
|
|
|
Post by johnofgwent on Jan 9, 2024 11:40:26 GMT
From my zoology undergraduate yesr
iQ or intelligence quotient is actually a test of one’s ability to act within one’s normal environment.
My grandfather and Alan Turing might be at the top of the ladder in the industrialised west but both score less than the average backwoods redneck hick if taken to the Serengeti and told to survive there.
The said backwoods redneck hick however would do remarkably well, though not as well as the local Masai tribesmen
|
|
|
|
Post by Orac on Jan 9, 2024 12:21:44 GMT
John, I think that concept is closer to the biological notion of 'fitness'. Intelligence can be a component of fitness but it isn't quite the same. For instance, Mr Turing may be able to get correct prescriptions from limited information, but is still unable to hack an attacker to death and so his intelligence searches, perhaps in vain, for another route through.
|
|
|
|
Post by Baron von Lotsov on Jan 9, 2024 15:06:38 GMT
I feel there is a lot of unnecessary faffing and fluffing about the exact definition of intelligence. IMHO Intelligence is the psychological ability to provide the one correct answer when an incorrect answer is far more likely. Accurate recall and retention is part of this ability, but the bit that seems to be most valuable is the part that creates an answer within a fixed (for all) information domain. This ability, if it is measured correctly, should be strongly linked to real world benefits for the holder. The acid test for IQ tests is that those with high measured IQs should in show more aggregate success in solving real life problems for themselves (ie have more success in getting what they want). This seems to be the case. Emotional intelligence almost certainly exists but the ability is a little tricky to capture objectively. The trouble is the brain itself has various characteristics. To get an answer to a problem some brains will give a fast and approximate answer, where other brains can ponder the problem for ages and come up with a very clever and better solution.
You also have to ask are problems in real life the binary sort, where there is one right answer and all the other possible answers are provably wrong. Brains tend not to work in this way anyway, and are more adapt at the successive approximation technique. If we look at maths and logic there are easy abstract puzzles to solve like three apples is greater than two apples and all the other simple maths you learnt at school. However in real life when dealing with physical systems, if you were to try and model their behaviour mathematically you would likely end up with a kind of partial differential equation that no known mathematician has been able to solve. So in nature the brain adopts the successive approximation method that although slower and does not give a precise answer, it solves much harder problems to an acceptable degree.
There are however brains that are better at fine detail and other brains which are better at assimilating the big picture, and these differences are to do with the neuron connections and likely a genetic trait. How are you going to weight these two different skills in your intelligence test, as per which has the higher value or is it better to sit in the middle.
Another trait of brains that somewhat buggers things up is brains have plasticity, so often you get a brain that specialises in one particular skill, and by doing so more of the brain's neurons lend themselves to this activity at the expense of things they could do and yet are not being used for.
I think the problem is the IQ test is a thing of the past where scientists really did not have any proper clue how the brain worked.
|
|
|
|
Post by Orac on Jan 9, 2024 15:22:15 GMT
Baron,
I don't think your objections are substantial. Any real world problem will have a set of 'answers' which are acceptably correct and a set that are unacceptably wrong. In this sense there is a binary in any problem with incorrect answers.
Any long set of problems will inadvertently test 'big picture' vs 'small picture' by giving the tested person the dilemma of spending more time on an intractable problem vs making an estimate and moving on.
|
|
|
|
Post by sandypine on Jan 9, 2024 15:34:55 GMT
Baron, I don't think your objections are substantial. Any real world problem will have a set of 'answers' which are acceptably correct and a set that are unacceptably wrong. In this sense there is a binary in any problem with incorrect answers. Any long set of problems will inadvertently test 'big picture' vs 'small picture' by giving the tested person the dilemma of spending more time on an intractable problem vs making an estimate and moving on. It seems the US army have the most experience and have set a limit of 80 I believe below which the army believes that a person is incapable of taking and following instruction effectively. It seems to work for the US army and they have been doing this and analysing this for over 100 years. |
|
|
|
Post by piglet on Jan 9, 2024 16:49:52 GMT
One definition of intelligence is being able to solve your own problems, whatever they are. Indeed, i wonder how many of you think that you dont have any, never mind doing something about it. Calculating payloads for rockets, long multiplication, or code for computers whatever they call it is nowt in comparison.
Yes you can change, if you are the same today as you were 50 years ago then somethings wrong. Einstein said doing the same thing over and over, and expecting a different result. One of my problems was that in younger years i thought every one was just like me.
HAAAAAAAAAAAAAA. You have a default setting, becoming aware of that and reacting to it should be no 1 priority, it would avoid so much suffering. But you havent, me neither, till i got divorced at 48, i made it my business to find out why it went wrong. All i can say is wow......completely different person now. Look and you will find, what i found is equally stunning, and shocking. Park your ego first.
|
|
|
|
Post by Baron von Lotsov on Jan 9, 2024 17:11:55 GMT
Baron, I don't think your objections are substantial. Any real world problem will have a set of 'answers' which are acceptably correct and a set that are unacceptably wrong. In this sense there is a binary in any problem with incorrect answers. Any long set of problems will inadvertently test 'big picture' vs 'small picture' by giving the tested person the dilemma of spending more time on an intractable problem vs making an estimate and moving on. It seems the US army have the most experience and have set a limit of 80 I believe below which the army believes that a person is incapable of taking and following instruction effectively. It seems to work for the US army and they have been doing this and analysing this for over 100 years. You could devise a test which tests the fitness to operate in the army and I'm more confident you would have better success with it than using a general intelligence test for measure of any job, not just a job in the army. The army is a relatively good fit for this kind of test because it works on a tall hierarchy, which is a system of management when you want tight control from the top down. Each cog in the system has a well-defined role and only has to carry out a small subset of all orders just centred around their narrow role. It is most ill-suited for selecting the person you want to devise your attack strategy. This is not about the skill of following orders, but a skill that draws on a lot of experience where there is no single correct answer, rather it is the analogue fuzzy world of better or worse within error bounds.
In fact it may well be the way the US army operates might have had an unduly large influence on wider corporate management strategy in that I understand the army funded a lot of research into psychology which end up as textbooks.
|
|
|
|
Post by Baron von Lotsov on Jan 9, 2024 17:29:44 GMT
Baron, I don't think your objections are substantial. Any real world problem will have a set of 'answers' which are acceptably correct and a set that are unacceptably wrong. In this sense there is a binary in any problem with incorrect answers. Any long set of problems will inadvertently test 'big picture' vs 'small picture' by giving the tested person the dilemma of spending more time on an intractable problem vs making an estimate and moving on. Are you aware of how ChatGPT answers a question?
It's not at all what you would think it is. When you type some text in, ChatGPT simply predicts what is most likely to come next, like sophisticated version of that autocomplete for typing in words, except it autocompletes a paragraph or even far more.
It follows that your answer will be dependent on how we ask the question. Looking at empirical research to do with surveys it is known that how you ask the question very much influences the answer, and also where you position the question. You are going to get some styles of phrasing the question more comprehensible to some brains and not other. For example mathematicians think in abstract symbols where most people like to think in terms of synthetic understanding of the world, like with numbers, many need to relate them to something they can picture.
|
|
|
|
Post by Orac on Jan 9, 2024 18:04:38 GMT
Baron, I don't think your objections are substantial. Any real world problem will have a set of 'answers' which are acceptably correct and a set that are unacceptably wrong. In this sense there is a binary in any problem with incorrect answers. Any long set of problems will inadvertently test 'big picture' vs 'small picture' by giving the tested person the dilemma of spending more time on an intractable problem vs making an estimate and moving on. Are you aware of how ChatGPT answers a question?
It's not at all what you would think it is. When you type some text in, ChatGPT simply predicts what is most likely to come next, like sophisticated version of that autocomplete for typing in words, except it autocompletes a paragraph or even far more.
It follows that your answer will be dependent on how we ask the question. Looking at empirical research to do with surveys it is known that how you ask the question very much influences the answer, and also where you position the question. You are going to get some styles of phrasing the question more comprehensible to some brains and not other. For example mathematicians think in abstract symbols where most people like to think in terms of synthetic understanding of the world, like with numbers, many need to relate them to something they can picture.
It seems if i attempt to discuss any of your points, you just go off on tangents. IQ tests may well have a verbal (language) component but usually this component is small (for a reason). I also have no idea at all why you feel your description of chat-gpt's operating mechanism doesn't fit my model of how it works |
|
|
|
Post by Baron von Lotsov on Jan 9, 2024 18:50:35 GMT
Are you aware of how ChatGPT answers a question?
It's not at all what you would think it is. When you type some text in, ChatGPT simply predicts what is most likely to come next, like sophisticated version of that autocomplete for typing in words, except it autocompletes a paragraph or even far more.
It follows that your answer will be dependent on how we ask the question. Looking at empirical research to do with surveys it is known that how you ask the question very much influences the answer, and also where you position the question. You are going to get some styles of phrasing the question more comprehensible to some brains and not other. For example mathematicians think in abstract symbols where most people like to think in terms of synthetic understanding of the world, like with numbers, many need to relate them to something they can picture.
It seems if i attempt to discuss any of your points, you just go off on tangents. IQ tests may well have a verbal (language) component but usually this component is small (for a reason). I also have no idea at all why you feel your description of chat-gpt's operating mechanism doesn't fit my model of how it works You don't tell us how you can devise a neutral set of questions. The problems are in which questions to ask and how you score the correctness of the answer, since if you select the questions to make these things easy then you are applying a bias. The point about Chat GTP was that it may well be doing something similar to the human brain. Indeed I find this problem in any test. What is the answer that would score you full marks and how detailed does it need to be. You get educationalists systemising knowledge. It's often the ones who use an alien system that are able to spot different answers since they look in different places (the search space of the solution). |
|
|
|
Post by Orac on Jan 9, 2024 19:28:30 GMT
It seems if i attempt to discuss any of your points, you just go off on tangents. IQ tests may well have a verbal (language) component but usually this component is small (for a reason). I also have no idea at all why you feel your description of chat-gpt's operating mechanism doesn't fit my model of how it works You don't tell us how you can devise a neutral set of questions. The problems are in which questions to ask and how you score the correctness of the answer, since if you select the questions to make these things easy then you are applying a bias. The point about Chat GTP was that it may well be doing something similar to the human brain. Indeed I find this problem in any test. What is the answer that would score you full marks and how detailed does it need to be. You get educationalists systemising knowledge. It's often the ones who use an alien system that are able to spot different answers since they look in different places (the search space of the solution). You haven't shown much in the way of substantial holes that require any explanation. The test questions tend to concentrate on human primitives - like shapes, 'colours', rotations and replacement and the tests have a clever internal self-check mechanism that splits the questions into type and makes the first question in the type set so easy that anyone who understood the quality of the question (what is being asked) would get the correct answer. Incorrect answers that follow this 'tutorial question' would show inability to deal with difficulty rather than not understanding the question. One might ask the same about the psychopathy checklist. The checklist works because people who score high tend to turn out to be psychopaths. With IQ tests and intelligence things are a bit firmer, in that people who score high tend to display the advantages one might expect from intelligence. Test familiarity is a known issue but doesn't seem to make that much odds (a few points). |
|
|
|
Post by Baron von Lotsov on Jan 9, 2024 19:56:51 GMT
You don't tell us how you can devise a neutral set of questions. The problems are in which questions to ask and how you score the correctness of the answer, since if you select the questions to make these things easy then you are applying a bias. The point about Chat GTP was that it may well be doing something similar to the human brain. Indeed I find this problem in any test. What is the answer that would score you full marks and how detailed does it need to be. You get educationalists systemising knowledge. It's often the ones who use an alien system that are able to spot different answers since they look in different places (the search space of the solution). You haven't shown much in the way of substantial holes that require any explanation. The test questions tend to concentrate on human primitives - like shapes, 'colours', rotations and replacement and the tests have a clever internal self-check mechanism that splits the questions into type and makes the first question in the type set so easy that anyone who understood the quality of the question (what is being asked) would get the correct answer. Incorrect answers that follow this 'tutorial question' would show inability to deal with difficulty rather than not understanding the question. One might ask the same about the psychopathy checklist. The checklist works because people who score high tend to turn out to be psychopaths. With IQ tests and intelligence things are a bit firmer, in that people who score high tend to display the advantages one might expect from intelligence. Test familiarity is a known issue but doesn't seem to make that much odds (a few points). The types of question are still only a small subset of all of skills the brain performs. They are abstract and logical plus somewhat detached from normal human work. For example, picture a mechanic stripping down an engine. This is far from a simple question, and it does involve a lot of intelligence. Humans have only realised exactly how much intelligence each task requires when they try and program a robot to do the job. Getting machines to do mathematical calculations like one gets on a calculator was accomplished a long time back, but looking in modern automated car factories where as much as possible is automated, you still have guys assembling most of the engine. Actually it has to be said that going back to the 1950s the popular idea was that soon computers would be able to do virtually anything. They massively underestimated what human intelligence was, and this was due to their simplistic style of thinking, using basic intuition. Still though the intelligence test is much the same as what it was back then. Me thinks these people are a little cocksure of themselves and have many convinced.
Anyway, if you want my view, I'd say you would do a lot better devising tests that test for a particular task you are employing the intelligence to do. You would get a much tighter feedback loop in perfecting your questions.
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Jan 9, 2024 23:34:24 GMT
Intelligence is a difficult thing to measure. Tests designed to measure it are always going to be somewhat arbitrary. And in any case our skills at negotiating such tests as individuals can vary according to a variety of factors, eg our levels of fatigue or tiredness, our levels of boredom, our emotional state.
Also, what exactly is intelligence? Some mistake memory for intelligence. Someone with a very good memory might know lots of bits of information but this itself is not a sign of actual intelligence. Though a more intelligent person would tend to make much better use of a good memory than a less intelligent one. Large amounts of knowledge about certain subjects would tend to require a certain level of intelligence but you hardly need to be a genius to have an encyclopaedic knowledge of football or opera, or film. You just need to have an interest in those fields, enough time to satisfy that interest by studying it, and a good enough memory to remember what you have learned, aided by the fact that we tend to find it much easier to memorise facts which interest us than we do facts which dont. But such knowledge itself does not really serve as a measure of intelligence, nor is it central to it. If intelligence were determined by knowledge, my knowledge of World War 2 would make me highly intelligent, and yet my lack of knowledge about football would make me highly unintelligent. So clearly, the possession of knowledge too, though made easier by intelligence, is not itself a reliable definition of intelligence either.
So again, what exactly is intelligence? I would say that it is in essence an ability to think independently, using logic and insight to draw conclusions from the facts you learn and the things you see and hear, and to be able to problem solve and figure out solutions for yourself. Also to recognise mistakes made by others and by yourself and to learn from them by figuring out alternative courses of action that might do more good and less harm. And to be able to figure out and predict likely outcomes to any action you have decided to take and think through all the possibilities you can logically determine based upon what you already know about the people and situations concerned.
So intelligence is not the ability to memorise lots of things. Nor is it the ability to acquire large amounts of knowledge about subject matter that is important to us, whether it be simply because we find it interesting, or because it is important in our jobs, or whatever. Intelligence is the ability when stumbling across a new situation, to assess it and everything around it, utilising anything in our memories that might be relevant, anything in our fields of knowledge that might be relevant, and to logically determine what actions to take, constructively using imagination combined with logic to figure out likely outcomes to any course of action.
Intelligent people will do this rather better than less intelligent ones.
Financial success however is not a failsafe guide to intelligence. You do not have to be smart to inherit a fortune, just the good luck to be born into the right family. I grant that to start off poor and become financially successful would require some degree of intelligence. But to reverse engineer that logical assumption and see a lack of financial success as a sign of lack of intelligence is patently untrue. Because it actually depends upon individual priorities. If achieving financial success with all the added pressures this can often bring is a primary goal, any intelligent person ought to be able to plan a course of action that can achieve this. Yet financial success is not the prime driving force for quite a lot of people. Enough to be secure and to be happy and to avoid too much pressure and be around for the family, whilst working in an emotionally rewarding environment, is a more important driver for some. And there are still people who pursue vocations as some sort of idealistic calling, as something they perceive as highly noble and therefore good for their own sense of self worth. Frequently this leads intelligent people onto career paths unlikely to bring them great financial success, eg nursing, teaching, or policing. And we would be in a sorry state indeed if it didn't.
|
|
