 |
|
|
 |
|
|
|
|
Ribald
[Guest] |
| | | | | | posted 11/3/2003 16:23 |  | | | | | | | | | | |
Nabarun Mondal wrote @ 11/3/2003 6:11:00 AM:
Exactly! Our brain does it Randomly, well almost randomly! The monitoring UTM randomly selects the exiting time. And our brain also does the same.Then comes again the question of Randomness in Intelligent beings...
| |
Nonsense.
It only seems random if you don't understand the underlying mechanism. Have you read any studies about human attempts at making random selections? We aren't any good at it. That most likely indicates that we don't function that way at lower levels.
You kind of missed the point with the Brownian Motion example. A random motion simulator is not at all the same thing as modeling Brownian Motion.
There have been plenty of random sentence generators, and they were clearly not as intelligent as Eliza.
The monotoring UTM would follow arbitrary rules, but rules nonetheless. We are talking about approximation, not about giving up on trying to be accurate.
| | |
| | |
Nabarun Mondal |
| | | | | | posted 11/4/2003 04:57 |  | | | | | | | | | | |
Ribald wrote @ 11/3/2003 4:23:00 PM:
Nonsense.
It only seems random if you don't understand the underlying mechanism. Have you read any studies about human attempts at making random selections? We aren't any good at it...
You kind of missed the point with the Brownian Motion example. A random motion simulator is not at all the same thing as modeling Brownian Motion.
There have been plenty of random sentence generators, and they were clearly not as intelligent as Eliza.....
| |
Then according to you , what is the underlying *mechanism* of Brownian motion, or electron difraction, or radioactivity, or laser...?
[1]As per as human mind is concerned, they can not choose randomly, because they *try to do it randomly* , or put in the more formal way *try to device an algo for generating random number and try ti stick to it*...and hence loose the path to randomness. What you may think of rules, are, and almost always statistical. If the nature *IS* inherently random, and follows almost in all the cases statistical rules, then we , natures creations , should not be thiniking ourselves extra-ordinary, should we?
[2]As per your underlying mechanism, have you read about reversible computation? It says that the more energy loose in computation , the more is the speed, and the more NON-RANDOM is your computation. In order to show this, for storing 1bit, our computers lose energy in the range exp(8) kT, where brain looses only 100kT. Which shows that the functionality of brain tends to be of random nature, as it has to be because, we have to accept the fact that there is no Intelligent thing on the back of our mind that is driving us! So, humen while thinking have to use Physical Laws, which are Statistical.
[3] As per as Eliza, it is not the kind of intelligence I am talking.Does Eliza misunderstands a word and let it be corrected, as a 2-3 year old baby does? No.
It may learn new rules, but that is not same as creating a new one. A random sequence generator is *not* intelligent, as electrons, protons, nutrons are not, not even C, N, O, H, P etc. But add them to form DNA, and some primitive intelligence (On the basis of which Genetic Algorithms are deviced) *IS* formed!
| | |
| | |
Ribald
[Guest] |
| | | | | | posted 11/4/2003 17:15 | | | | | | | | | | | |
Nabarun Mondal wrote @ 11/4/2003 4:57:00 AM:
[0] Then according to you , what is the underlying *mechanism* of Brownian motion, or electron difraction, or radioactivity, or laser...?
[1]As per as human mind is concerned, they can not choose randomly, because they *try to do it randomly* , or put in the more formal way *try to device an algo for generating random number and try ti stick to it*...and hence loose the path to randomness. What you may think of rules, are, and almost always statistical. If the nature *IS* inherently random, and follows almost in all the cases statistical rules, then we , natures creations , should not be thiniking ourselves extra-ordinary, should we?
[2]As per your underlying mechanism, have you read about reversible computation? It says that the more energy loose in computation , the more is the speed, and the more NON-RANDOM is your computation. In order to show this, for storing 1bit, our computers lose energy in the range exp(8) kT, where brain looses only 100kT. Which shows that the functionality of brain tends to be of random nature, as it has to be because, we have to accept the fact that there is no Intelligent thing on the back of our mind that is driving us! So, humen while thinking have to use Physical Laws, which are Statistical.
[3] As per as Eliza, it is not the kind of intelligence I am talking.Does Eliza misunderstands a word and let it be corrected, as a 2-3 year old baby does? No.
It may learn new rules, but that is not same as creating a new one. A random sequence generator is *not* intelligent, as electrons, protons, nutrons are not, not even C, N, O, H, P etc. But add them to form DNA, and some primitive intelligence (On the basis of which Genetic Algorithms are deviced) *IS* formed!
| |
[0] You need to read up on Brownian Motion, Einstein himself said it better than I can. The motion of the particle is caused by the molecules in the suspension interacting with the particle. This is a good example because there were all kinds of outlandish theories (including "animation") to explain a behavior that is causal and logical once you understand the interactions that they could not see with the devices of the time. The behavior seemed "random", which is why some attributed life to it.
We tend to attribute anthropomorphic characteristics to things that happen for reasons that we do not understand. That is a human psychological trait, which I would hope intelligent people try to overcome. Unless we want to go back to worshipping river gods and such.
[1] You seem to be confused about statistical analysis and causality. Just because statistics can indicate trends or probabilities does not mean that the underlying causality is "random", just that it is complex. You can have a statistical probability which is extremely high, yet get the opposite results in any given specific case. That is why psychology and sociology are not formal sciences. I really do not understand what you mean by "statistical rules". There is no such thing, there is only statistical probabilities that can be attributed to complex systems when you do not have a complete model of al of the variables.
[2] That sounds like specious reasoning to me. The efficiency of a processing system does not determine whether or not it behaves randomly, you may have drawn an incorrect conclusion from something you read. There is no such thing as a random computation, only pseudo-random. You can base algorithmic behavior upon external variables, which makes the result SEEM random, but it still isn't. You need to separate math and procedure, this is what I tried to address originally.
[3] Eliza versus a random sentence generator was an example to help you see that "randomness" is not intelligence. Intelligent creatures act on their own internal reasoning, and are therefore not completely predictable to us, which is why we tend to attribute intelligence to behaviors we do not understand. That is a logical fallacy. DNA is not intelligent, we just don't understand it completely yet. I don't think that my car is intelligent or alive, but someone from 200 years ago probably would.
| | |
| | |
Rob Hoogers |
| | | | | | posted 11/5/2003 10:42 | | | | | | | | | | If I may say a word in Naharun's defense:
randomness *is* an important factor in intelligent behaviour. Impredictability can be a success factor, and there are such things as 'feleticious errors', basically embarrassingly stupid decisions that turned out gold, for reasons only later understood clearly. This is also 'semi-random' behaviour in the sense that it is unpredictable for an observer who expects rational decisions only.
The last winners in such bot contests switched to randomness whenever they perceived a succesful counterstrategy applied to their previous actions. It's a brute-force effort, but highly successful.
| | | Last edited by Rob Hoogers @ 11/5/2003 10:43:00 AM | | |
| | |
Ribald
[Guest] |
| | | | | | posted 11/5/2003 16:25 | | | | | | | | | | | |
Rob Hoogers wrote @ 11/5/2003 10:42:00 AM:
If I may say a word in Naharun's defense:
randomness *is* an important factor in intelligent behaviour. Impredictability can be a success factor, and there are such things as 'feleticious errors', basically embarrassingly stupid decisions that turned out gold, for reasons only later understood clearly. This is also 'semi-random' behaviour in the sense that it is unpredictable for an observer who expects rational decisions only.
The last winners in such bot contests switched to randomness whenever they perceived a succesful counterstrategy applied to their previous actions. It's a brute-force effort, but highly successful.
| |
Rob, did you actually read the discussion?
Predictability to external (or even internal!) observers has nothing to do with the subject at hand.
Your final paragraph merely reinforces the fact that we attribute intelligence to unpredictability. Why do we do that? Because intelligent beings do things for internal reasons that we do not understand, therefore it is a perhaps instinctual response to assume animation in complex behavior.
"randomness *is* an important factor in intelligent behaviour."
First off, I don't believe that there is such a thing as "intelligent behavior". There is behavior that may be exhibited for intelligent reasons, but that is not the same thing. We could go back to the old "Martian Box" discussion, but let's not unless we have to.
But let me re-frame your statement, I would posit that a truer form would be that unpredictability is important to our PERCEPTION of underlying intelligence.
I do not see anything you have said that really supports his argument. And I don't think he needs defense...
| | |
| | |
Rob Hoogers |
| | | | | | posted 11/5/2003 19:03 | | | | | | | | | | | |
Ribald wrote @ 11/5/2003 4:25:00 PM:
[1] Rob, did you actually read the discussion?
[2] I do not see anything you have said that really supports his argument. And I don't think he needs defense...
| |
[1] Ofcourse not, this was just a random remark.
("Then comes again the question of Randomness in Intelligent beings..." -previous page -)
;)
[2] Unpredictability can be a strategy. The maximum effect is reached where the opponent cannot distinguish between your next move and a random one.
This is not the same as something only in the eye of the observer. It can be deliberate, or caused by erroneous behaviour, which can be inpredictable, again. Sun Tzu said it, Milosevic does it, to the despair of his judges.
So, even when humans cannot be random in a true sense, like a bot or a program, we can and do come close.
If this is done on purpose, there can only be one conclusion, at least to the intelligent observer.... ;)
| | |
| | |
Ribald
[Guest] |
| | | | | | posted 11/5/2003 19:13 | | | | | | | | | | | |
Rob Hoogers wrote @ 11/5/2003 7:03:00 PM:
[2] Unpredictability can be a strategy. The maximum effect is reached where the opponent cannot distinguish between your next move and a random one.
| |
Ah, yes, I quite agree with that.
That was not the source of debate, but is a valid assertion I think.
Part of a successful higher predator is the ability to predict behavior.
| | |
| | |
Rob Hoogers |
| | | | | | posted 11/6/2003 09:17 | | | | | | | | | | Which brings us back to the main issue: we want to be surprised, wowed and amazed by people, and mutatis mutandis also by an AI.
As Raphael remarked in the Turing Revisited thread, some people tend to become more predictable in behaviour these days, making the goal of AI easier if anything (my rephrase).
I think it would actually be quite possible to come up with humans that would fail the Turing. We'd balk collectively as these humans would not be 'representative' of us as a species. I doubt that. I think they make up a major part of the population at any given time.
In short, we have people who can come up with the most boring and uninteresting, even simpleton conversation.
Look at the threads in this forum, where young people are most impressed by simpleton, cursing and sex-crazed bots, simply because it 'speaks their language'. For them they are the most convincing
bots around. They don't want to talk to a bot that sounds like their demented uncle.
In essence we're trying to create something near-superhuman, while we dare not judge our fellow humans by those same standards.
Most bots will do nicely for a lot of people, and only the best is good enough for a few.
So the problem with AI is we're partly hunting a mirage. Those few will up the ante for AI time and again, because they also learn more about themselves, the human animal. And we do evolve ourselves by doing this. But as usual the evolution is unevenly distributed.
| | | Last edited by Rob Hoogers @ 11/6/2003 9:21:00 AM | | |
| | |
Nabarun Mondal |
| | | | | | posted 11/6/2003 13:03 | | | | | | | | | | | |
Rob Hoogers wrote @ 11/6/2003 9:17:00 AM:
Which brings us back to the main issue: we want to be surprised, wowed and amazed by people, and mutatis mutandis also by an AI......
| |
This is what we can call *Information* in AI.
[ high information contain is something with low prob of occurence]
Pure (read Uniform ) randomness is never Intelligent and Pure Logic is not what we actually intended. We want things that can do something that we don't tell it to do, and sometime which will be predictable, and sometimes not. That will be AI, in it's truest sense...and that is what anything that we call intelligent does!
But , then as uniform distributions does not stick to any laws, they are truly random, but not Gaussian one's they are predictable in nature... and also Poisson distbn. These may be a piller for AI, as the Fuzzy logic incorporates some randomness if not all in decision making.
| | |
| | |
Rob Hoogers |
| | | | | | posted 11/11/2003 14:25 | | | | | | | | | | | |
Ribald wrote @ 11/4/2003 5:15:00 PM:
[0] You need to read up on Brownian Motion,
| |
[0] So I did, so I did.
Ilya Prigogine's La fin des Certitudes (1996) to be exact. Interesting book, that. Couldn't find any translation into English, but it would probably translate as "The End of Certainties", or somesuch. Brilliant stuff.
Discusses the arrow of time, brownian motion, chaotic behaviour, the works in short.
He points out that the individual behaviour of particles and the statistical behaviour of particles, in a diffusing gas for instance, show us some behaviour that was heretofore not seen in the right perspective: in a Hilbert phase space these ensembles of possible statistical and individual behaviour are expressed as frequencies, with resonance influencing each other. At the point where these resonances diverge, calculation becomes impossible, and classical determinism breaks down. This puts the relation between statistical and individual behaviour into a completely new light.
Also things like swarm behaviour/individual behaviour can be expressed in such a fashion, opening new avenues into seeing consciousness properly.
More, his discussion of a diffusing gas set off something else in my head, namely that 'ordinary' particles and larger process are *all* involved in a diffusing process, namely the expansion of the universe itself. We've been talking about in on big scales, because that's where we first encountered it.
But it's going on at the smallest level too, of course. And that's a kind of process that is at one level invisible to us, but could be the explanation for what makes the whole gamut of physical processes possible in the first place, and maybe the irreversibility of Time itself, since the expansion of the Universe is a one-way process, as far as we know.
Maybe this silly hubble expansion is what makes Change *possible*, physically, and literally.
BTW wasn't Brownian motion a favourite subject of someone else at this forum, too? ;)
| | |
| | |
Nabarun Mondal |
| | | | | | posted 11/12/2003 12:18 | | | | | | | | | | | |
Rob Hoogers wrote @ 11/11/2003 2:25:00 PM:
More, his discussion of a diffusing gas set off something else in my head, namely that 'ordinary' particles and larger process are *all* involved in a diffusing process, namely the expansion of the universe itself. We've been talking about in on big scales, because that's where we first encountered it.
But it's going on at the smallest level too, of course. And that's a kind of process that is at one level invisible to us, but could be the explanation for what makes the whole gamut of physical processes possible in the first place, and maybe the irreversibility of Time itself, since the expansion of the Universe is a one-way process, as far as we know.
Maybe this silly hubble expansion is what makes Change *possible*, physically, and literally.
BTW wasn't Brownian motion a favourite subject of someone else at this forum, too? ;)
| |
Will someone please go to the Einsteins equation for Brownian motion?
If somebody sees that , eventually she/he will find a classic correspondence of it with statistical equations, and from there it can also be found that (possibly) the expansion of anything with some temparature is one of the statistical thing possible for it to do!
| | |
| | |
Pennywise |
| | | | | | posted 11/12/2003 17:45 | | | | | | | | | | | |
Nabarun Mondal wrote @ 11/12/2003 12:18:00 PM:
Will someone please go to the Einsteins equation for Brownian motion?
| |
Perhaps I'm misunderstanding you, but here is Einstein's explanation of Brownian motion in a nice little java applet:
http://www.phys.virginia.edu/classes/109N/more_stuff/Applets/brownian/brownian.html
| | |
| | |
Sam Fentress |
| | | | | | posted 11/13/2003 02:15 | | | | | | | | | | | |
Nabarun Mondal wrote @ 11/12/2003 12:18:00 PM:
Will someone please go to the Einsteins equation for Brownian motion?
If somebody sees that , eventually she/he will find a classic correspondence of it with statistical equations, and from there it can also be found that (possibly) the expansion of anything with some temparature is one of the statistical thing possible for it to do!
| |
Yes and no.
Brownian motion is certainly derived from statistical laws (or rather, even if brownian motion can't be properly described by classical mechanics and statistics, let's pretend it can for the sake of argument). But does this mean that it's random? No. The exact motion of each particle will still follow it's own deterministic path (QM doesn't play a particularly large part here, I believe).
Let's have a different example. If a million people were selected, we'd expect to have roughly half men and half women. But as you say, this is a matter of statistics. It's remotely possible that everyone selected will be men. But does it follow that the sex of each person is random? No, it's completely deterministic based on what chromosomes they were given by their parents.
So yes, it's statistically possible that all the air in the room you are in will suddenly bunch up in one corner. But it doesn't follow from this that Brownian motion is random.
| | |
| | |
Rob Hoogers |
| | | | | | posted 11/13/2003 17:47 | | | | | | | | | | To be precise about it: George Gamow said that you could calculate the *probability* of gas bunching up in one corner, never that it followed that it actually would... It's a probability, 's all.
Brownian motion is causal, but unpredictable, deterministic chaos. Representation as a random system can be easier at times, more practical even, but it is not random in the classical sense.
| | |
| | |
Pennywise |
| | | | | | posted 11/13/2003 21:40 | | | | | | | | | | | |
Rob Hoogers wrote @ 11/13/2003 5:47:00 PM:
Brownian motion is causal, but unpredictable, deterministic chaos.
| |
It's predictable if you can capture a whole state of the system. This is Hard and currently technically infesible, but not impossible.
Not that you really said anything wrong, I just wouldn't want people confusing deterministic chaos with randomness.
| | |
| | |
Ribald
[Guest] |
| | | | | | posted 11/14/2003 20:41 | | | | | | | | | | | |
Pennywise wrote @ 11/13/2003 9:40:00 PM:
Not that you really said anything wrong, I just wouldn't want people confusing deterministic chaos with randomness.
| |
Yes, precisely. I think that is where some smart people go astray in their thinking, by confusing complexity with randomness.
I think this often is the erroneous conclusion they come to after failing to find a clear linear representation via reduction.
In addition to considering "Brownian Motion", I think it is illuminating to read some of Wolfram's work on complexity and reduction. I believe that his CA work is of great importance to understanding a massively parallel structure like the brain.
I think the below should be required reading in computer science. For one thing it shows the fallacy of Penrose's (and others') statements about UTM.
http://www.stephenwolfram.com/publications/articles/ca/88-complex/2/text.html
| | |  http://www.stephenwolfram.com/publications/articles/ca/88-complex/2/text.html | | |
| | |
Nabarun Mondal |
| | | | | | posted 11/24/2003 06:06 | | | | | | | | | | | |
Sam Fentress wrote @ 11/13/2003 2:15:00 AM:
Let's have a different example. If a million people were selected, we'd expect to have roughly half men and half women. But as you say, this is a matter of statistics. It's remotely possible that everyone selected will be men. But does it follow that the sex of each person is random? No, it's completely deterministic based on what chromosomes they were given by their parents....
| |
Good. Then the selection of chromosome... is it not random? Can humen *control* X & Y chromosomes? [ Forget artificial pregnancy, think about the natural ways...]
The root will always remain nondeterministic....
| | |
| | |
Rob Hoogers |
| | | | | | posted 11/24/2003 11:08 | | | | | | | | | | | |
Pennywise wrote @ 11/13/2003 9:40:00 PM:
It's predictable if you can capture a whole state of the system. This is Hard and currently technically infesible, but not impossible.
| |
Knowing *all* variables in a system to the penultimate decimal is impossible, and you know it.
Unless you think you're something else. ;)
| | |
| | |
Ribald
[Guest] |
| | | | | | posted 11/24/2003 16:02 | | | | | | | | | | | |
Rob Hoogers wrote @ 11/24/2003 11:08:00 AM:
Knowing *all* variables in a system to the penultimate decimal is impossible, and you know it.
Unless you think you're something else. ;)
| |
That depends upon the complexity of the system, doesn't it?
There is a difference between your ability to predict something and whether or not it is a deterministic event.
Randomness means non-deterministic, yes?
That is not the same as something that you personally do not know the outcome of ahead of time.
What if two people are observing an event, one knows the outcome in advance, the other does not. Is it a random event or not?
Logic would dictate surely not, unless you are a solipsist.
| | |
| | |
Rob Hoogers |
| | | | | | posted 11/24/2003 19:38 | | | | | | | | | | | |
Ribald wrote @ 11/24/2003 4:02:00 PM:
That depends upon the complexity of the system, doesn't it?
| |
Yup. Starts with the three-body problem already. Not too complex, until you start adding decimals of precision and for every decimal added you get a wildly different result for your calculation.
So 'complex' is a station easily passed without even noticing it.
But already in this case, to be able to predict anything at all would require knowledge to the nth decimal, which is *infinite*. And since we're not the Supreme Being, we just *can not tell*.
Once the moving bodies have assumed their respective positions, we can ofcourse always backtrack our ways in a deterministic manner. How else could they've arrived there... ;)
| | | Last edited by Rob Hoogers @ 11/24/2003 7:44:00 PM | | |
|
|
|
|
|
