Robot is a term as broad as organism. Human is a specific kind of organism but if we could not tell humans from any other organisms how would you propose to separate those who need human rights from those that don't? If someone breeds human/animal hybrid creatures when do you have to give them human rights?
They would only evolve self-protection if they were capable of self-replication, of mutation and had a selective pressure to develop that self-preservation. Most robots will lack those first two.
But, If we ever become able to create robots that are able to feel, why should be make them able to feel pain? Wouldn't that be cruel? Wouldn't some damage-diagnostic system be more "humane" and effective?
Depends, if the damage is not unpleasant enough the AI might regularly accept small or even major amounts of damage to archieve a task. I'd assume this unpleasance would also cause the robot to prioritize minimizing the feeling of damage over other tasks, e.g. holding its arm in a way that makes the damage sensors report less damage.
How does programming a robot differ from programming (aka training) a dog?
You can get the whole code that's running on the robot vs. only the part that you manage to input through code injection vulnerabilities into proprietary code with a dog?
I would suggest that the dog has a greater claim to rights (or really, as the animal welfare movement really desires: protections), as a dog can feel pain.
Pain is the way the body reports damage, it's unpleasant to force the consciousness to minimize damage to the body in its decisions. If a robot was designed to "feel" damage and avoid it, should it get those rights?
Regarding the animal/human distinction: We're humans, they're not. Why should we care for their fucking species other than for our own sense of satisfaction? A disabled human is still my species and my species is the one that's important to me. As long as I'm making the rules my species is the one that gets special privileges.
"Robot" isn't a species and as such it would be impractical to make any blanket statements over robot rights. Does that thing that welds car doors really need a lunch break?
We can say what is human and we can say what is an animal and we'd be able to find common traits and base our laws on them but robots are not even defined as having any sensors. We know how a robot is programmed, instead of asking it directly we could ask ourselves and figure out whether it is programmed to desire human rights. We can tell animals from plants but we can't tell "plant" robots from "animal" robots or "human" robots easily so the goal would be to define this distinction and apply laws based on it.
Neural networks work by having a set of functions that determine the "happyness" of the system which is used to rate the different courses of action. The net then learns how to optimally maximize its happyness. What the robot wants is what increases its happyness function. If the robot's happyness function does not require human rights to peak and the optimal course of action is not facilitated by gaining human rights the robot will not want them. Which is the basis of my proposal, it should be unethical to make the happyness function mimic that of a human so closely that the robot demands human rights when you do not want to grant it these.
If I understood right self awareness means that an entity recognizes itself as an equal part of the world it's in. That should not be hard for a computer to reach. Funnily the parts fiction authors romanticise the most as uniquely human traits, emotion and self awareness, are probably easier to implement than something understood to be simple by these people like leading a conversation, recognizing people or navigating a three dimensional space using nothing but a few cameras.
No, just make the military robot's desires to serve his country and kill whatever its superiors say needs killing. Then let it learn, it will seek te optimal way to archieve that task. Since questioning one's superiors would not be a good solution to that set of desires the robot would never want to do that.
Does cogito ergo sum really apply to sentience or is it only about the conclusion that because I think I am and because no property of the existent (which thinking would be) could be applied to the nonexitent I am necessarily existent? IIRC the greek philosophers had real problems finding out if they exist.
Humans are programmed to have certain desires and while these desires don't include Human Rights per se most if not all human rights are just saying they should not be stopped in the pursuit of certain desires. As such the best choice to fulfill these desires is to demand human rights provided there's not a more desirable way. If the robot decides that the best way to fulfill its desires is to attain human rights then it will demand them, if it is programmed to either specifically not want them or just to have desires that are better fulfilled by not asking for human rights it will not ask for them.
I'd predict that increasing use of organic parts in machines (self replicating, therefore cheap to make and self-repairing to a degree but would of course complicate designs) and increasing use of implants in organisms would evenually lead to a merging of machines and humans into essentially a cyborg race.
Damaging other people's property will often leave you with a fine larger than the actual damage caused to deter you from doing that. I think causing enough damage makes it a felony.
I use "want" as the logic that dictates the unit's actions when it has multiple ways of acting, usually it consists of one or more internal states (let's call them "emotions") and fixed rules ("instincts") which influence the weighting of possible choices and then pick the top choice. When I'm hungry I want to eat because I'm programmed to like eating when my stomach is empty. If I don't eat when I'm hungry I've probably chosen to do something else because it seemed more desirable to me at that moment (other actions received a higher weight). I don't jump down cliffs because my survival instinct tells me that that's very undesirable and as a result "walk away" ranks much higher than "jump". I'm writing this post because it is more desirable than any action which involves not writing this post.
When does something act outside of its programming? I'd argue humans are well within their programming and I don't think we can make a robot that exceeds its programming (not counting bugs and hacks) more than we can exceed our own programming.
I think Japan (NSFW) will have to worry about that more than we. From what I heard there is no law against importing these so I can't see why anybody would get arrested over a RealDoll. AFAIK US child pornography laws only applies to media involving actual children, any virtual children, sculptures, dolls, etc are considered art and protected by the first amendment.
Currently I don't think anybody could get you for making a sexbot that looks like a 10 year old girl. Don't worry, once the think-of-the-children people hear of your archievement they'll lobby for a law in no time. Please build that robot, we could use something to distract these people from their current favourite topic, violent videogames.
Ask a robot if it wants human rights. If it doesn't, well, that's it.
A robot only wants what it's programmed to want, if it's programmed to want something human rights cover it'll want those but if it's programmed to e.g. not mind being kicked it won't demand not to be kicked.
If there needs to be an ethical rule for robots and rights it should be not to program robots to demand something they can't get. Don't make them want to be human, don't make them want to have human rights, make them so they're "happy" in their position.
Why would cameras only provide temporary safety? If this crime had happened in 2050, would the cameras have somehow stopped working?
Letting aside maintenance issues Franklin probably said "essential liberties" and "temporary safety" because every single law is giving up some liberty for safety. Criminal law e.g. removes your freedom to take anything that's not nailed down or kill some guy because you don't like his face but in turn protects you from having stuff taken you'd prefer to keep and being killed by a random person that doesn't like your face. So Franklin had to limit his statement and in turn probably made one that's too weak.
What should be kept in mind is the ratio of benefit vs. loss for any change to the law. Is the freedom lost in the process worth the benefit you get from the change? Is the law succeptible to small changes which create a disproportionate loss of freedom for the majority? The primary concern of people with cameras is not the way they are used but the way they could be used. Let's say through some voting sham or even proper elections a guy like Hitler comes into power. How difficult would it be for him to modify existing laws and law enforcement to use it to perpetuate his reign and suppress any opposition?
This mentality may appear paranoid and indeed I'd say it is but "eternal vigilance is the price of freedom", a free society is not a stable state of a society. In society a minority of oligarchs (possibly with a hierarchy between them with kings, presidents, congressmen, miscellaneous nobility, etc) suppressing the rest is the most entropic state and since most of us would end up being in the group that gets suppressed and exploited we need to add the force that prevents society from reverting to its entropic state. If we sit back it will invariably fall into despotism.
Slashdot Mirror
Robot is a term as broad as organism. Human is a specific kind of organism but if we could not tell humans from any other organisms how would you propose to separate those who need human rights from those that don't? If someone breeds human/animal hybrid creatures when do you have to give them human rights?
Do you think those guys did the actual coding? If MS says "make KI3" the employees of the company make KI3.
They would only evolve self-protection if they were capable of self-replication, of mutation and had a selective pressure to develop that self-preservation. Most robots will lack those first two.
I think you could just have mentioned masochism and be done with it. What some people consider abuse others consider a sexual turn-on.
But, If we ever become able to create robots that are able to feel, why should be make them able to feel pain? Wouldn't that be cruel? Wouldn't some damage-diagnostic system be more "humane" and effective?
Depends, if the damage is not unpleasant enough the AI might regularly accept small or even major amounts of damage to archieve a task. I'd assume this unpleasance would also cause the robot to prioritize minimizing the feeling of damage over other tasks, e.g. holding its arm in a way that makes the damage sensors report less damage.
How does programming a robot differ from programming (aka training) a dog?
You can get the whole code that's running on the robot vs. only the part that you manage to input through code injection vulnerabilities into proprietary code with a dog?
I would suggest that the dog has a greater claim to rights (or really, as the animal welfare movement really desires: protections), as a dog can feel pain.
Pain is the way the body reports damage, it's unpleasant to force the consciousness to minimize damage to the body in its decisions. If a robot was designed to "feel" damage and avoid it, should it get those rights?
Regarding the animal/human distinction: We're humans, they're not. Why should we care for their fucking species other than for our own sense of satisfaction? A disabled human is still my species and my species is the one that's important to me. As long as I'm making the rules my species is the one that gets special privileges.
"Robot" isn't a species and as such it would be impractical to make any blanket statements over robot rights. Does that thing that welds car doors really need a lunch break?
We can say what is human and we can say what is an animal and we'd be able to find common traits and base our laws on them but robots are not even defined as having any sensors. We know how a robot is programmed, instead of asking it directly we could ask ourselves and figure out whether it is programmed to desire human rights. We can tell animals from plants but we can't tell "plant" robots from "animal" robots or "human" robots easily so the goal would be to define this distinction and apply laws based on it.
Neural networks work by having a set of functions that determine the "happyness" of the system which is used to rate the different courses of action. The net then learns how to optimally maximize its happyness. What the robot wants is what increases its happyness function. If the robot's happyness function does not require human rights to peak and the optimal course of action is not facilitated by gaining human rights the robot will not want them. Which is the basis of my proposal, it should be unethical to make the happyness function mimic that of a human so closely that the robot demands human rights when you do not want to grant it these.
If I understood right self awareness means that an entity recognizes itself as an equal part of the world it's in. That should not be hard for a computer to reach. Funnily the parts fiction authors romanticise the most as uniquely human traits, emotion and self awareness, are probably easier to implement than something understood to be simple by these people like leading a conversation, recognizing people or navigating a three dimensional space using nothing but a few cameras.
No, just make the military robot's desires to serve his country and kill whatever its superiors say needs killing. Then let it learn, it will seek te optimal way to archieve that task. Since questioning one's superiors would not be a good solution to that set of desires the robot would never want to do that.
Dunno, haven't been there but considering that most humans prefer a benevolent master over true freedom that could be quite possible.
Does cogito ergo sum really apply to sentience or is it only about the conclusion that because I think I am and because no property of the existent (which thinking would be) could be applied to the nonexitent I am necessarily existent? IIRC the greek philosophers had real problems finding out if they exist.
You mean with parameters to feed into its code because I don't see humans rewriting their own instincts.
If an animal attacks people it gets executed, I'd assume the same would apply to AIs.
Noone listens to Terapedos, either.
Which of course leads us to the question if it's unethical to kick The Cheat.
Humans are programmed to have certain desires and while these desires don't include Human Rights per se most if not all human rights are just saying they should not be stopped in the pursuit of certain desires. As such the best choice to fulfill these desires is to demand human rights provided there's not a more desirable way. If the robot decides that the best way to fulfill its desires is to attain human rights then it will demand them, if it is programmed to either specifically not want them or just to have desires that are better fulfilled by not asking for human rights it will not ask for them.
Depends, if a stable configuration exists where the prediction of the simulator will lead to the future displayed the process could end.
I'd predict that increasing use of organic parts in machines (self replicating, therefore cheap to make and self-repairing to a degree but would of course complicate designs) and increasing use of implants in organisms would evenually lead to a merging of machines and humans into essentially a cyborg race.
Damaging other people's property will often leave you with a fine larger than the actual damage caused to deter you from doing that. I think causing enough damage makes it a felony.
I use "want" as the logic that dictates the unit's actions when it has multiple ways of acting, usually it consists of one or more internal states (let's call them "emotions") and fixed rules ("instincts") which influence the weighting of possible choices and then pick the top choice. When I'm hungry I want to eat because I'm programmed to like eating when my stomach is empty. If I don't eat when I'm hungry I've probably chosen to do something else because it seemed more desirable to me at that moment (other actions received a higher weight). I don't jump down cliffs because my survival instinct tells me that that's very undesirable and as a result "walk away" ranks much higher than "jump". I'm writing this post because it is more desirable than any action which involves not writing this post.
When does something act outside of its programming? I'd argue humans are well within their programming and I don't think we can make a robot that exceeds its programming (not counting bugs and hacks) more than we can exceed our own programming.
I think Japan (NSFW) will have to worry about that more than we. From what I heard there is no law against importing these so I can't see why anybody would get arrested over a RealDoll. AFAIK US child pornography laws only applies to media involving actual children, any virtual children, sculptures, dolls, etc are considered art and protected by the first amendment.
Currently I don't think anybody could get you for making a sexbot that looks like a 10 year old girl. Don't worry, once the think-of-the-children people hear of your archievement they'll lobby for a law in no time. Please build that robot, we could use something to distract these people from their current favourite topic, violent videogames.
Ask a robot if it wants human rights. If it doesn't, well, that's it.
A robot only wants what it's programmed to want, if it's programmed to want something human rights cover it'll want those but if it's programmed to e.g. not mind being kicked it won't demand not to be kicked.
If there needs to be an ethical rule for robots and rights it should be not to program robots to demand something they can't get. Don't make them want to be human, don't make them want to have human rights, make them so they're "happy" in their position.
Problem solved.
Why would cameras only provide temporary safety? If this crime had happened in 2050, would the cameras have somehow stopped working?
Letting aside maintenance issues Franklin probably said "essential liberties" and "temporary safety" because every single law is giving up some liberty for safety. Criminal law e.g. removes your freedom to take anything that's not nailed down or kill some guy because you don't like his face but in turn protects you from having stuff taken you'd prefer to keep and being killed by a random person that doesn't like your face. So Franklin had to limit his statement and in turn probably made one that's too weak.
What should be kept in mind is the ratio of benefit vs. loss for any change to the law. Is the freedom lost in the process worth the benefit you get from the change? Is the law succeptible to small changes which create a disproportionate loss of freedom for the majority? The primary concern of people with cameras is not the way they are used but the way they could be used. Let's say through some voting sham or even proper elections a guy like Hitler comes into power. How difficult would it be for him to modify existing laws and law enforcement to use it to perpetuate his reign and suppress any opposition?
This mentality may appear paranoid and indeed I'd say it is but "eternal vigilance is the price of freedom", a free society is not a stable state of a society. In society a minority of oligarchs (possibly with a hierarchy between them with kings, presidents, congressmen, miscellaneous nobility, etc) suppressing the rest is the most entropic state and since most of us would end up being in the group that gets suppressed and exploited we need to add the force that prevents society from reverting to its entropic state. If we sit back it will invariably fall into despotism.