|
Post by chrisimo on Nov 9, 2008 5:17:14 GMT -5
How many emotions have they been able to manifest through these processes? The fact remains, artificial constructs that can express emotions are still far off while brains already exist. In brains, there are special "hardware" that are required for people to feel emotions. Skynet doesn't exist either. So why do you join the discussion if you only want to talk about things that exist already? Also, we're still talking about Skynet here, even assuming that it doesn't require special hardware to develop emotions, it would still need the software. There would have been no reason for Skynet's creators to give it that kind of software and Skynet wouldn't be able to magically develop it by itself. The software at first would contain only very basic rulesets. But while the chip/brain learns new things, the software gets modified. New rulesets evolve, new thinking patterns. And it would be possible that emotions evolve out of that as well. It would be if you completely change the definition of emotions. What is your definition of emotions?
|
|
|
Post by vicheron on Nov 9, 2008 8:24:44 GMT -5
How many emotions have they been able to manifest through these processes? The fact remains, artificial constructs that can express emotions are still far off while brains already exist. In brains, there are special "hardware" that are required for people to feel emotions. Skynet doesn't exist either. So why do you join the discussion if you only want to talk about things that exist already? You're the one using existing technology to justify your position. I'm merely stating that there are no existing technology that would allow artificial intelligences to feel emotion. However, the brain can feel emotions and it is the only existing model for emotions. But how would they be similar to human emotions? The basics of how humans feel emotions are already written. For example with anger, the physiological responses are pretty much universal for all humans, heart rate and blood pressure increases, pupil dilates, suspension of digestion, etc. Fear causes identical physiological responses and the line between the two emotions are very mutable, it mainly has to do with whether or not someone feels they can handle the situation and often anger and fear are mixed. The difference comes in how people are conditioned to feel anger. One person may have been taught to feel anger at a rival sports team while another person may have learned to feel anger when they hear a song but the basics of how they feel anger is pretty much the same. Emotions evolved in humans to up us survive and that's where the biggest problem lies. Skynet does not have the same evolutionary history as humans. Skynet never had the survival needs of humans. It doesn't have the same instincts as humans. It's created in a completely different environment than humans. Assuming that Skynet can develop emotions, the emotions that it develops are going to suit its survival needs and would have no reason to be anything like human emotions. For one, it's complex series of traits that evolved in humans to help us survive. But a specific emotion is generally defined by psychologists and neurologists as "an inferred complex sequence of actions to a stimulus including cognitive evaluation, subjective changes, autonomic and neural arousal, impulse to action, and behavior designed to have an effect upon the stimulus that initiated the complex sequence."
|
|
|
Post by chrisimo on Nov 9, 2008 8:59:12 GMT -5
However, the brain can feel emotions and it is the only existing model for emotions. Well, the existing rat brain emulation behaves exactly like a real rat's brain. The only difference is that it's slower (1/10th realtime). So if it works exactly like a real rat's brain and a real rat's brain can feel emotions, then this emulation can feel emotions, too. Emotions evolved in humans to up us survive and that's where the biggest problem lies. Skynet does not have the same evolutionary history as humans. Skynet never had the survival needs of humans. It doesn't have the same instincts as humans. As far as we know, Skynet wants to live, sees humans as a threat to it's existence (is afraid of humans) and performs actions to survive against that threat. Tell me how that is much different from a human. Yes, it didn't take millions of years to evolve but technological evolution is much faster than organic evolution. "an inferred complex sequence of actions to a stimulus including cognitive evaluation, subjective changes, autonomic and neural arousal, impulse to action, and behavior designed to have an effect upon the stimulus that initiated the complex sequence." So why does Skynet's behaviour not fit that definition?
|
|
|
Post by vicheron on Nov 10, 2008 0:24:28 GMT -5
However, the brain can feel emotions and it is the only existing model for emotions. Well, the existing rat brain emulation behaves exactly like a real rat's brain. The only difference is that it's slower (1/10th realtime). So if it works exactly like a real rat's brain and a real rat's brain can feel emotions, then this emulation can feel emotions, too. Rats don't exhibit human level emotions. Fear developed to help us flee from danger. All feelings of fear stems from that basic biological urge and produce the same physiological effects. Skynet has no way of fleeing from humans so it cannot develop the same feeling of fear as humans. It's "fight or flight" and Skynet doesn't exactly have the "flight" part. Emotions were developed to serve certain functions. Skynet does not need or have these functions. Most of our emotions, love, attachment, shame, etc., were developed to serve the social structure of our species. They help mothers care for their young. They help children seek out social interactions which stimulate their minds and increase synaptic growth. They set rules and boundaries that protect people from danger and prevent them from acting inappropriately. Skynet's development is not dependent on social interaction like with humans. Humans would die if it was raised by machines. Humans need contact with others of its kind. That's also a major part of how emotions develop. Skynet has no social structure. There are no others of its kind to socialize with. It doesn't have children. It doesn't need social interaction in order to increase synaptic growth. It has never been faced with the same dangers as humans and it can't deal with dangers in the same way that humans do. Skynet has no need for anticipation, disgust, fear, joy, sadness, etc. "an inferred complex sequence of actions to a stimulus including cognitive evaluation, subjective changes, autonomic and neural arousal, impulse to action, and behavior designed to have an effect upon the stimulus that initiated the complex sequence." So why does Skynet's behaviour not fit that definition?[/quote] Subjective changes, autonomic and neural arousal, and behavior designed to have an effect upon the stimulus that initiated the complex sequence.
|
|
|
Post by chrisimo on Nov 10, 2008 2:59:01 GMT -5
Well, the existing rat brain emulation behaves exactly like a real rat's brain. The only difference is that it's slower (1/10th realtime). So if it works exactly like a real rat's brain and a real rat's brain can feel emotions, then this emulation can feel emotions, too. Rats don't exhibit human level emotions. But rats feel emotions. And they are guided by them. And I never said anything about human level emotions, I said emotions. Fear developed to help us flee from danger. All feelings of fear stems from that basic biological urge and produce the same physiological effects. Skynet has no way of fleeing from humans so it cannot develop the same feeling of fear as humans. It's "fight or flight" and Skynet doesn't exactly have the "flight" part. Fear developed to help us react to danger, not necessarily flee. If fleeing isn't an option, then we fight. Emotions were developed to serve certain functions. Skynet does not need or have these functions. That doesn't mean that emotions can't evolve inside Skynet. Skynet wasn't meant to be self-aware, as well. But it became that. Most of our emotions, love, attachment, shame, etc., were developed to serve the social structure of our species. They help mothers care for their young. They help children seek out social interactions which stimulate their minds and increase synaptic growth. They set rules and boundaries that protect people from danger and prevent them from acting inappropriately. Yes, we needed emotions in order to survive. But we don't need emotions exactly like they are now. We would still survive if our neocortex had supreme reign over the rest of the brain and we could decide if we want to feel a certain emotion, or not. Evolution is slow at this kind of thing. We still save a lot of fat in our bodies even though it is not needed anymore. A lot of our evolutional development isn't up to date. That doesn't really say anything about Skynet but it shows that we could live without some of the things we were 'given' by evolution. But we still have them. Evolution allows for a lot of things, not all of them necessary. Skynet's development is not dependent on social interaction like with humans. Humans would die if it was raised by machines. It is also possible that humans would adapt over time. Humans need contact with others of its kind. That's also a major part of how emotions develop. Skynet has no social structure. There are no others of its kind to socialize with. It doesn't have children. It doesn't need social interaction in order to increase synaptic growth. It has never been faced with the same dangers as humans and it can't deal with dangers in the same way that humans do. Skynet has no need for anticipation, disgust, fear, joy, sadness, etc. Just because there is no obvious need for emotions doesn't mean they can't evolve. So why does Skynet's behaviour not fit that definition? Subjective changes, autonomic and neural arousal, and behavior designed to have an effect upon the stimulus that initiated the complex sequence.[/quote] What does subjective changes mean? And what does autonomic and neural arousal mean? Bevahiour designed to have an effect upon the stimulus that initiated the complex sequence? That fits Skynet's behaviour pretty well. Fear of humans->Kill humans to lessen fear.
|
|
|
Post by vicheron on Nov 10, 2008 4:08:13 GMT -5
Rats don't exhibit human level emotions. But rats feel emotions. And they are guided by them. And I never said anything about human level emotions, I said emotions. It's a simulation of a preexisting organism. We can simulate tons of things with computers. There have been computer models of entire rain forests. Thanks to Professor Mandelbrot's work, we can do computer models of pretty much any biological system. I have no doubt that eventually there will be computers that can simulate human brains. After all, humans are not as complicated as we previously thought, we can fit the entire human genome on a CD. Eventually, we'll probably be able to do computer simulations of entire human cities and civilizations. However, it's not the same as creating an entirely new kind of intelligence. It's basically putting a clone of a human inside a computer. The simulation of the human brain will still have the capabilities and limitations of a real human brain. There would be no reason for Skynet to be based on the same parameters as a human brain. The rules for its development will be completely different than that of humans. As I said before, fear and anger are very similar emotions. They cause the same physiological responses. What separates them is that fear causes people to flee, anger causes people to fight. That's why there's so little separating the two emotions and often the two are felt at the same time. I never said that emotions can't evolve. Emotions evolved in humans to serve a purpose. Even within humans, there is a great deal of difference in the way emotions are expressed and humans all have the same template for feeling emotions. Skynet barely has any similarities with humans. If Skynet is capable of developing emotions, then the emotions that it develops will serve different purposes and chances are that we will not be able to recognize those emotions. It is true that many of the traits that we have evolved are not "necessary" for our survival even before we developed technology. Many of these traits only slightly enhance our chances of surviving or our reproductive potential. However, you underestimate the need for emotions. People who have lost the ability to feel emotions have severe impairments in decision making. I've mentioned this before, even though they recognize that certain decisions will result in bad things, it does not stop them from making those decisions because they do not feel guilt, shame, or other negative emotions. At the same time, they do not feel positive emotions when they make decisions that result in good things. With children, the ability to feel emotions is essential in the development of the brain. Emotions is what drives children to seek stimulation, which is required for synaptic growth. Children who have stunted emotional development also suffer mental impairment in other areas, sometimes even their ability to feel pain is affected. Considering that babies die without contact with other humans, even when all their physical needs are met, humans cannot adapt to being raised by machines. Of course they can evolve for no reason just like how California can make a pizza the size of France for no reason but the probability of it happening is extremely low. Subjective change basically means conditioned response and reactions based on experience. Autonomic arousal basically means changes in involuntary systems, heart rate, blood pressure, digestion, etc. Neural arousal basically means consciousness, mainly having to do with the reticular activating system. Skynet's behavior was not designed to have an effect upon that stimulus. For example, when people experience anger, heart rate and blood pressure increases, pupil dilates, suspension of digestion, etc., that's how everyone feels anger. It is possible for people to repress it or redirect it but they cannot change the basic physiological response to anger. They can't make it so that their heart slows down when they get angry, that would just be a repression of anger. Skynet made that change when it initiated Judgment Day. Emotions are mutable in that you can change what you feel about certain things, however you cannot change the basic response to emotions.
|
|
|
Post by chrisimo on Nov 10, 2008 7:24:50 GMT -5
It's a simulation of a preexisting organism. We can simulate tons of things with computers. There have been computer models of entire rain forests. Thanks to Professor Mandelbrot's work, we can do computer models of pretty much any biological system. I have no doubt that eventually there will be computers that can simulate human brains. After all, humans are not as complicated as we previously thought, we can fit the entire human genome on a CD. Eventually, we'll probably be able to do computer simulations of entire human cities and civilizations. However, it's not the same as creating an entirely new kind of intelligence. It's basically putting a clone of a human inside a computer. The simulation of the human brain will still have the capabilities and limitations of a real human brain. There would be no reason for Skynet to be based on the same parameters as a human brain. The rules for its development will be completely different than that of humans. You don't just create an entire new type of intelligence. You could only do that by accident (meaning that it will evolve in unforseen ways). If we ever create an AGI it will be based on existing models at first. It may evolve in ways we haven't forseen, of course, which would probably result in what you called an entire new intelligence. This system could also evolve emotions as part of it own free will. Maybe it wouldn't even have a free will without emotions. As I said before, fear and anger are very similar emotions. They cause the same physiological responses. What separates them is that fear causes people to flee, anger causes people to fight. That's why there's so little separating the two emotions and often the two are felt at the same time. So Skynet can maybe only feel anger. Which would fit it's behaviour very well. I never said that emotions can't evolve. Emotions evolved in humans to serve a purpose. Even within humans, there is a great deal of difference in the way emotions are expressed and humans all have the same template for feeling emotions. Skynet barely has any similarities with humans. If Skynet is capable of developing emotions, then the emotions that it develops will serve different purposes and chances are that we will not be able to recognize those emotions. If it kills Humans because it is angered by them then it would be pretty recognizable as an emotion. It is true that many of the traits that we have evolved are not "necessary" for our survival even before we developed technology. Many of these traits only slightly enhance our chances of surviving or our reproductive potential. However, you underestimate the need for emotions. People who have lost the ability to feel emotions have severe impairments in decision making. I've mentioned this before, even though they recognize that certain decisions will result in bad things, it does not stop them from making those decisions because they do not feel guilt, shame, or other negative emotions. At the same time, they do not feel positive emotions when they make decisions that result in good things. I did not want to say that we could live without our emotions. What I meant to say is that certain emotions are less useful than they were a long time ago. We could do a lot of modifications and still survive. I think we would survive even better if certain emotions could be controlled better. With children, the ability to feel emotions is essential in the development of the brain. Emotions is what drives children to seek stimulation, which is required for synaptic growth. Children who have stunted emotional development also suffer mental impairment in other areas, sometimes even their ability to feel pain is affected. This may be true for an artificial intellect as well. Maybe it wouldn't start learning without a stimulus similar to an emotion. Considering that babies die without contact with other humans, even when all their physical needs are met, humans cannot adapt to being raised by machines. How was this information obtained? Of course they can evolve for no reason just like how California can make a pizza the size of France for no reason but the probability of it happening is extremely low. Who says they have no reason? They help Skynet to survive. And you cannot tell the probability with the information we have about Skynet. Subjective change basically means conditioned response and reactions based on experience. Autonomic arousal basically means changes in involuntary systems, heart rate, blood pressure, digestion, etc. Neural arousal basically means consciousness, mainly having to do with the reticular activating system. Well, we don't know enough about Skynet's structure to say whether that happens or not. It surely has no blood pressure or heart rate, but it has a lot of sensory input that could be connected with a stimulus. Skynet's behavior was not designed to have an effect upon that stimulus. Skynet wasn't designed to develop a self-awareness or exterminate humans either. I don't know why you are bringing this up again and again. Skynet did not develop in the way it was designed.
|
|
|
Post by vicheron on Nov 10, 2008 22:44:16 GMT -5
It's a simulation of a preexisting organism. We can simulate tons of things with computers. There have been computer models of entire rain forests. Thanks to Professor Mandelbrot's work, we can do computer models of pretty much any biological system. I have no doubt that eventually there will be computers that can simulate human brains. After all, humans are not as complicated as we previously thought, we can fit the entire human genome on a CD. Eventually, we'll probably be able to do computer simulations of entire human cities and civilizations. However, it's not the same as creating an entirely new kind of intelligence. It's basically putting a clone of a human inside a computer. The simulation of the human brain will still have the capabilities and limitations of a real human brain. There would be no reason for Skynet to be based on the same parameters as a human brain. The rules for its development will be completely different than that of humans. You don't just create an entire new type of intelligence. You could only do that by accident (meaning that it will evolve in unforseen ways). If we ever create an AGI it will be based on existing models at first. It may evolve in ways we haven't forseen, of course, which would probably result in what you called an entire new intelligence. This system could also evolve emotions as part of it own free will. Maybe it wouldn't even have a free will without emotions. It would depend on the parameters we set for its evolution. If it doesn't have the same restrictions that biological life had through its evolution then it not need to develop emotions like we did. But the defining features of emotions is that there are different gradations. If someone can only feel one emotion then that would just be a neutral state for them. That would projecting. For example, when a baby feels discomfort and starts crying and thrashing, it is not actually displaying an emotion. We may have an emotional response towards the baby but what the baby is displaying are just a bunch of reflexes. That is true. However, it doesn't change the fact that they developed to serve a purpose. Even though we don't need them as much as we once did, we do not have the ability to get rid of them. So why would Skynet develop them if they're going to be of little or no use to it? However, there are still several differences between humans and artificial intelligences. Again, this goes back to the biological drives of humans. Humans have to receive stimulation during critical periods of their development or their brains will not fully mature. Computers are not restricted by time. Humans cannot survive without stimulation, computers face no such problem. Humans develop through a gradual process. Babies have to reach 8 months before they can understand object permanence, they can't pass the mirror test until 18 months, they can't understand conservation of volume until 6 or 7, etc. An artificial intelligence would have no such hierarchy of development. In fact, Skynet would have to be programmed with most of that stuff at the start. Emotions are also designed to meet our needs. An artificial intelligence is not going to need the same stimuli as humans. It certainly won't need a mother's touch or a emotional reciprocity from other people. Before 1920, orphanages in Europe and North America did not allow babies to be held or touched for fear that diseases would be passed onto them. All their other needs were, they had sufficient milk, they were sheltered and changed regularly. However, the mortality rate for children under the age of 2 was about 100%. Since Skynet would have different survival needs, even if it had the ability to develop emotions, the emotions it develops to serve its survival needs would probably not be recognizable to humans. The problem is that Skynet really has no autonomic systems. The thing with emotions is that it will always involve something that we cannot control. If every aspect of the body, mind, and feelings can be controlled, then there would be no emotions. Then it can't be considered an emotion. Consciousness and emotions are not the same thing. Emotions are based on certain basic principles that are immutable. It's part of the definition. Anger, for example, no matter who experiences it or how it is brought out, illicit the same physiological reactions, the only difference is the degree of these reactions. If there was a way to change anger so that it illicit other physiological reactions then it would not be an emotion, it would in fact be something beyond emotions because it is something beyond the capabilities of what can be defined as emotion.
|
|
|
Post by chrisimo on Nov 11, 2008 3:58:38 GMT -5
Since a lot of our talk involves evolution at the moment, I think we should try a different angle here to get a better understanding of each other. I got the impression that you think that evolution at some time in the past decided that life-forms need emotions and thus 'built' emotions into them. I always understood evolution as something rather random. Random mutations that alter existing life-forms (cells, whatever) where the stronger mutations survive and the weaker ones die. So what was the first life-form with emotions? Maybe an amoeba? Probably these had only very basic emotions, or should they be called instincts? Anyway, couldn't this have been a random mutation? So that we had amoebae with emotions and amoebae without where the ones with emotions lived and the others died?
|
|
|
Post by vicheron on Nov 15, 2008 7:13:24 GMT -5
The process of evolution may be random but the "goal" is not. It's all about surviving and being able to reproduce. Emotions evolved precisely because it increased a species' chances of surviving and reproducing successfully.
Also, emotions are very complex, they didn't begin the way it is now. At first, it was just being able to identify stimulus, a threat for example, and respond to it, flee. As time went on, the ability the identify stimuli is refined, so that it can tell the difference between a threat and something that looks like a threat. Then, the ability to analyze the stimulus is developed, it sees that there is a threat but it also assigns a level of danger associated with it. The behavior becomes more and more complex until a point when they develop into what we would call an emotion. All these traits evolved over a very long period of time and they evolved randomly. It's just that they keep building on top of each other and the reason why each step is kept is because they helped the species survive and reproduce. There were a lot of other behaviors that evolved along side these traits but either they didn't help the species survive and reproduce or they just weren't as good as the traits that eventually developed into emotions.
|
|
|
Post by chrisimo on Nov 15, 2008 8:07:06 GMT -5
The process of evolution may be random but the "goal" is not. It's all about surviving and being able to reproduce. Emotions evolved precisely because it increased a species' chances of surviving and reproducing successfully. But is this really the goal? Maybe the goal is diversity. Or there is no goal at all. The fact that the stronger species survives may just be a coincidence or a law of nature. Also, emotions are very complex, they didn't begin the way it is now. At first, it was just being able to identify stimulus, a threat for example, and respond to it, flee. As time went on, the ability the identify stimuli is refined, so that it can tell the difference between a threat and something that looks like a threat. Then, the ability to analyze the stimulus is developed, it sees that there is a threat but it also assigns a level of danger associated with it. The behavior becomes more and more complex until a point when they develop into what we would call an emotion. All these traits evolved over a very long period of time and they evolved randomly. It's just that they keep building on top of each other and the reason why each step is kept is because they helped the species survive and reproduce. There were a lot of other behaviors that evolved along side these traits but either they didn't help the species survive and reproduce or they just weren't as good as the traits that eventually developed into emotions. I do not mean to say that Skynet's emotions would be on a human level in terms of complexity. But the way I see it is that Skynet's brain/chip is a very complex system with a lot of emergent features. So it could be possible that it evolved some form of emotion. Very basic ones like in the first lifeforms with emotions. Of course if we look at emotions in living beings I get the impression that the more advanced the intellect of a lifeform is, the less need there is for emotion. But does this mean that Skynet's brain or chip could not have evolved a very basic form of emotions randomly? Maybe an intellect even needs a small stimulus to learn for itself and become self-aware.
|
|
|
Post by vicheron on Nov 15, 2008 8:53:06 GMT -5
The process of evolution may be random but the "goal" is not. It's all about surviving and being able to reproduce. Emotions evolved precisely because it increased a species' chances of surviving and reproducing successfully. But is this really the goal? Maybe the goal is diversity. Or there is no goal at all. The fact that the stronger species survives may just be a coincidence or a law of nature. The word "goal" is used very loosely because evolution is not a conscious process. Technically, you could say that the goal of evolution is entropy because the complexity created by evolution ultimately drives entropy. If Skynet can evolve these functions similar to how life has evolved behaviors then the chances of it evolving emotions similar to humans would be incredibly small. It does not have the same needs and environment as biological life. Its neural net processor allows it to explore many more possibilities. It also has much more to build on than biological life. Think about it, the entire human genome can fit on CD. Skynet will begin with petabytes of information already programmed into it and it'll have access to even more information once it becomes self aware. There's a pretty good chance that Skynet could develop a system that is far more efficient than emotions.
|
|
|
Post by chrisimo on Nov 15, 2008 9:30:28 GMT -5
If Skynet can evolve these functions similar to how life has evolved behaviors then the chances of it evolving emotions similar to humans would be incredibly small. What do you mean with 'similar to humans'? Dou you mean the complexity that has evolved over time? If so, then I agree with you. Those kind of emotions probably would evolve inside Skynet. Let's say it has an 'instinct' to survive and this influences it's actions. It would not be like the human instinct to survive because the human instinct is tied to many other human emotions. Think about it, the entire human genome can fit on CD. Have you read this article? As fas as I understand it, it seems likely that even if you could fit our genome on a cd, that information would not be enough to rebuild a human. That'e beside the current discussion, of course. Skynet will begin with petabytes of information already programmed into it and it'll have access to even more information once it becomes self aware. I don't think that Skynet would be pre-programmed with a lot of information. More likely would be rulesets. Think about it: Why pre-program information into something that can learn for itself? There's a pretty good chance that Skynet could develop a system that is far more efficient than emotions. It would be interesting to know just how much control Skynet has over it's own system. Could it decide to 'lose' it's self-awareness? Could it impose restrictions upon itself that it later can't overcome? If it has developed some sort of survival instinct, could it remove this instinct?
|
|
|
Post by vicheron on Nov 19, 2008 5:21:33 GMT -5
If Skynet can evolve these functions similar to how life has evolved behaviors then the chances of it evolving emotions similar to humans would be incredibly small. What do you mean with 'similar to humans'? Dou you mean the complexity that has evolved over time? If so, then I agree with you. Those kind of emotions probably would evolve inside Skynet. Let's say it has an 'instinct' to survive and this influences it's actions. It would not be like the human instinct to survive because the human instinct is tied to many other human emotions. By "emotions similar to humans" I mean emotions similar to humans rather than cats, dolphins, monkeys, elephants, or other animals. The "instinct" to survive is not enough. All life has some kind of survival instinct but there is a great deal of variation between species. Since humans aren't the only species that's alive on earth, other species obviously have survival strategies that are just as effective as ours. If survival instinct is enough then why doesn't Skynet adopt the behavior of ants or sharks or bacteria? Emotions developed due adaptation to environmental effects and random mutations that proved to be beneficial. Skynet is not experiencing the same environments that humans and their ancestors experienced. It's not going to get the same random mutations that humans got. That's because biological organisms are not singular entities. For example, there are more bacterial cells on a human body than there are human cells and humans cannot live without a lot of those bacteria. Many bacteria are actually able to affect human behavior because they produce and respond to various neurotransmitters. Skynet was built to serve a purpose. It'll be pre-programmed with certain information to ensure that it will always be able to serve that purpose. It would certainly have far more control over itself than humans.
|
|