well it is hard to tell without predicting the long term future. I mean if we eradicate all malaria something else may also happen as a consecuence that we were not aware now but is bad for us.
Until we become intelligent enough to get our resources without hurting any other species on the planet, we'll keep defending our "own right" to life against anything else, and just try to minimize the casualties on the other "side".
Until we manage that, as the supreme species on the planet, we will get to arbitrarily decide what gets to live and what doesn't. All the other life forms are at our mercy. At least most of them aren't even aware of it.
Well, we certainly don't do that with animals. Only after a certain threshold of intelligence we don't want to hurt/eat them, or if they are "wild life", which we generally want to preserve, although it's only in recent history that this has happened.
If aliens existed, and came here, I wonder if they would treat us the same depending on whether they are only 2x, 10x or 1000x more intelligent than we are. Of course it's also possible that the more intelligent they get, the more they will want to preserve any kind of life, and will try to get their resources from something else to survive.
It should get interesting if we ever develop human brain-like artificial intelligence, and then we amplify it by 1000x following Moore's law for a couple of decades. Kill switches would become useless. We'd just have to teach them to value life more than anything, before it's too late.
But why would it be? If you give it the first law, then that command should be "disabled" for anyone from outside. You could try to tell a robot that, but it shouldn't work. Your best chance should be if you try to trick the robot into doing a certain action that would indirectly get that person dead. but if he's smart enough to know how to protect a human life, he should be smart enough to deny you the command, knowing it will endanger that other person's life.
In 2003 there were 6 million car accidents, not to mention people falling over, wars, etc. You're mandating the robot to put everyone in a padded cell for their own good.
it is important to mention "ability to survive by its own or in group" to be politically correct, for example little babies usually cant survive on its own but can survive in a group. so they usually have right to live. morality is just another way of saying i care for this baby so i form a group with this baby.
Perhaps this is an argument in favor of some sort of perverted natural selection, where you have a right to life commensurate with your ability to defend it?
something like that, anyway im not saying that i know the future of machine and know it would survive. and therfore i don't know if it has right to live. saying that something doesn't have right to live is not same as being able to kill it or shut it down.
i mean if some machine is not usefull for nothing and it wastes resources than you have right to shut it down if you can, no matter what IQ it has.
I think it will take quite some time for humans to accept robots as entities with rights. There's a lot of fear surrounding the idea of intelligent machines, and most deeply religious people (who I know, family included) consider robots/A.I. to be (somewhat) evil. We mostly have Hollywood to thank for that.
When the majority of the human population understands technology and its ever-changing limits, the fear and stigma towards artificial intelligence comparable to our own will subside and only then will sentient robots be accepted.
It's human nature to fear the unknown. It's gotten us to where we are today. But I think there comes a point in our timeline where we overcome all fear with our ability to understand and predict. Ironically enough, this will probably require the advancement of (and potentially cooperation with) A.I.
I think it's pretty likely that people will say that very complete, high-level AI (if it is ever invented) would have a right to life. Actually, there's a fantastic exploration of this from an AI's point of view in Life Artificial by David A. Eubanks:
Even if the robots themselves won't have many rights, humans will try to get them more rights by electing the politicians that will do it. Why would humans do that? Because I believe we will become emotional towards them. We can easily become emotional towards our pets, so it should be even easier with a robot that is even smarter than that. Heck, some people are even somewhat emotional towards their iPhones.
The first step towards becoming more emotional will be naming them. I remember reading a couple of years ago about that cleaning robot Roomba, and how people started naming them, and when they were giving them away for repairing, they demanded the same robot back, not a replacement.
I kinda agree with you but then, there's the uncanny valley. I think slavery ended by slaves fighting for their rights more than the 'masters' being the source behind the movement. I read many 'masters' were emotionally and even romantically involved but didn't fought so much as cope with the system.
You raise an interesting point with the slavery. I'm not sure if it's such a good analogy, but saying it is, Abraham Lincoln did try to free the slaves. There will be people, probably a lot of people, who will feel threatened by them, just like now there are people being afraid of robots taking their jobs (which is another interesting debate), but things will still change gradually, as robots become smart enough, "lovable" enough, and human-like enough, and some people will even want to marry them.
A few years ago I was reading sort of a joke, that by 2050, Massachusetts will be the first state that will allow human-to-robot marriages.
weren't humans put to death by killing or stealing horse. when horse was vital resource this was considered righteous thing to do. So to answer;. if AI was somehow vital resource to someone , it could possibly be that someone would get sentenced to death if one disables it.
If some countries that have capital punishment don't get civilized by the time we have human-level AI, then yes, that would be logical. There is no difference between a mind run on a biological brain and a silicon brain. Killing either is murder.
A truly intelligent and self-aware computer will be created by simply simulating a human brain on a powerful computer. I can't see how that mind would have any less rights than its biological counterparts.
So the answer is: Yes, it'll have the exact same rights, as we do.
Morally, that is, because the law might take time to catch up.
If the simulation decides to copy itself, and allows the copy to run independently, are there now two autonomous people?
If the copy is allowed to advance one nanosecond and is then deleted, leaving the original running, has a murder been committed? What if the copy never runs at all? What if the original is deleted instead of the copy?
What if the copy receives identical input, and thus has identical state as the original, up to the point it is deleted? What if some aspects of the copy are "optimized" by reusing the results of the original? What if the entire copy just mirrors the state of the original, without recomputing anything? How can we say exactly how many copies actually exist?
What if the original simulation modifies the copy so that it wants to kill itself, which it does, once it starts running? What if it's modified to just not care whether it lives or dies? What if the copy is modified in various other ways? Are some kinds of modifications ethical, while others are not?
Point being, this course of AI development does not spare us having to completely rethink our models of ethics.
Very interesting questions, I'll be happy to try to answer them:
1) yes
2a) yes
2b) It never lived, it was a "frozen corpse", and since you can make copies at will, just having a copy doesn't constitute a new life until it's run.
2c) If the original has been deleted at the same state as the copy is started, i.e. there's no interruption from the subjective point of view of the copy, then it's just a transfer.
3a) Since there's zero information difference, I'd say it is OK.
3b) They're two separate people with equal rights.
3c,d) No computing = no thinking = no life to speak of.
4a,b) The same as if you kidnapped someone from the street and convinced them to kill themselves (say, by drugging them).
4c,d) Very interesting. On the one hand, in case of many modifications, you can't really foresee the consequences, so all should be considered unethical to prevent unintended suffering.
On the other hand - what is human parenting and bringing up if not shaping (or "modifying") a not-yet-sovereign mind. And we have no problem with that. But then, what if some psychotic parent brings up a kid with an implicit conviction that he should kill himself upon reaching adulthood (i.e. legal sovereignty). We'd put them for life in jail. But what if a parent has no bad intentions, but is just a very crappy parent and his kid grows up to be very miserable and in his 20's commits suicide because he can't deal with life. Well, we could pass moral judgment on the parent (if we somehow had the knowledge and thus certainty that it was a result of bad bringing up), but we couldn't really convict them in a court of law.
So yeah, we're having a serious split here between moral judgment and what's practical. (Much like pro- and anti- abortion.)
So I'd say playing with the brains of sentient beings is unethical in general. Practically though, there's no way to stop a "virtual" person from running all kinds of crazy stuff on their private machines. Unless you want to police the internals of all such brain-simulating computers, but then you're conducting mind-control of people, and that's totally unacceptable.
5) Agreed, but I think if we engage in it already, we can get to some very interesting conclusions that we might find useful even in today's world.
2b) It never lived, it was a "frozen corpse", and since you can make copies at will, just having a copy doesn't constitute a new life until it's run.
How long does it have to run before it's alive? If the interval is short enough then there will be no change in state whatsoever. So in what way does the mind state have to change in order for it to have lived? What if that change is so simple that you can calculate it in your head? Have you created a new life just by thinking about what will happen next in the simulation? By writing it down?
3a) Since there's zero information difference, I'd say it is OK.
So you would have no objection to being killed if there was an identical copy of you in existence somewhere? Why does that matter? Why do you care more about this copy of you than you do about anybody else? Why should a stranger care that there is at least one copy of you?
4c,d) Very interesting. On the one hand, in case of many modifications, you can't really foresee the consequences, so all should be considered unethical to prevent unintended suffering.
But you are modifying your own mind all the time, by forming memories and learning skills. To some degree, you choose what you think and experience. Imagine that you have ultimate control over every facet of your mind. You live in a computer, so there is no "natural" mental life. You must choose exactly how your mind will progress.
If you make a frozen copy of yourself, it hasn't diverged yet, so you can modify it however you like, since it's you.
And if new, unique people are to be created at all, someone will have to choose how their mind works. There is no longer a "natural" reproductive process to decide this. How can choices be made in the new person's interests when it is those very interests that are being chosen?
2b) I would assume that the machine on which the AI is run is digital (i.e. discrete), so yeah, a single cycle (tick of the processor) could very well mean change. And since you've got to draw the line somewhere, I'd say that's enough.
3a) By killing me and starting an identical copy of me somewhere else, you're creating a discontinuity that infringes upon my identity (or same-ness) because there would be a very strong discontinuity in inputs simply because of the different location.
Moreover, a third party exercising control over any of my copies (and especially the one that is running) is in violation of my sovereignty. [That's also why a stranger should care about my copies.]
4c,d) That's precisely it: I am doing it! I can do with myself whatever I want. And yes, I do imagine having the control about every facet of my mind. I actually dream about it. :)
Now, when I manipulate a copy of me, that's a little different, because as I've stated earlier, a copy spun off to run independently is sovereign from its "original". That's why manipulating own copy should morally be considered equivalent to manipulating a third party, assuming that copy will be let run separately.
To your last point, I agree, that's the most delicate part - creating new beings. As far as copies are concerned, my immediate instinct is that only identical copies should be considered ethical, since any modifications can have unpredictable consequences. But I'm not sure this should be enforced by the AI community / data center, because I value personal sovereignty and habeas corpus virtualis above all else. ;)
I won't speak of creating completely new, original beings, because that topic deserves a ton of thought and a philosophical treatise of its own. But theoretically, you could try to mimic natural processes of child brain development. Of course, as you've pointed out, someone has to choose the parameters of that developmental algorithm, just as a search engine is never really objective, because people wrote it and made some choices in the process. But that last example also shows we can have some trust in such choices for practical reasons.
On a side note: I would highly recommend the book "Diaspora" by Greg Egan. The first chapter (Orphanogenesis) speaks about the development of a virtual mind and to me is among the best works of science fiction of all time.
You know the dating site OKCupid? It matches people by their answers to user-submitted questions.
One of about a dozen questions on there I've marked as "important to me" is one on AI rights. And interestingly enough, I'd say the site's suggestions have become markedly better since I did that; I think that question is a barometer for a lot of things about one's morals and interests.
Isn't that the most likely result? There are a lot of possible things we could call intelligent, and only a few of them could be "like us".
(Granted, anyone trying to make an AI will aim towards something like us, but considering how we still haven't developed AI yet, I imagine the first successful one will be more of a lucky accidentally-on-purpose than anything strictly-designed; even with humans making it, the first success will be quite non-human.)
An intelligent machine has the distinct advantage that its entire consciousness is duplicable and transferrable. You could turn the machine off and on again and it would be just as 'alive' as if it had never been switched off. I think the idea that human life is sacred is largely predicated on the fact that mortality is permanent, so it may not apply to AI in the same way. Perhaps a more salient idea would be "right to regular and secure backups."
Of course, this brings up the troubling thought experiment: if you could copy your own brain, and install it in a new body after your death, would that still be you? Would that satisfy your "right to life"? Our present theory of consciousness is insufficient to answer this question conclusively. My opinion is that consciousness is epiphenomenal.
Rather than answer this right now, I would suggest that everyone go watch a season or two of Battlestar Galactica, which probes this question deeply. Cylons (the "skin job" kind) were severely subjected to a kind of anti-machine racism, even though they were basically indistinguishable from humans.
There will probably be at least 4 stages, with the condition that their intelligence will not be limited, and it will keep growing at least until it's equal to ours:
1) No rights. We can do whatever we want, just like they are our property and we can treat them as pure objects
2) Some rights. Think of it like animal rights, but we still own them
3) "Official" human-like rights. They are free, but there would be "racism" against them at first
4) The exact same rights as a human, officially and unofficially.
Then things might start to change again when they start becoming drastically more intelligent than us. Then it becomes much harder to predict. At first glance, we could say it will be bad for us, but I choose to believe the optimistic side that the smarter they will be the more tolerant they will be, too.
Yes, definitely. Unless it bore a passing resemblance to an Atari game, in which case Atari and Apple would drag it outside and put a bullet in its head.
46 comments
[ 3.7 ms ] story [ 106 ms ] threadUntil we manage that, as the supreme species on the planet, we will get to arbitrarily decide what gets to live and what doesn't. All the other life forms are at our mercy. At least most of them aren't even aware of it.
If aliens existed, and came here, I wonder if they would treat us the same depending on whether they are only 2x, 10x or 1000x more intelligent than we are. Of course it's also possible that the more intelligent they get, the more they will want to preserve any kind of life, and will try to get their resources from something else to survive.
It should get interesting if we ever develop human brain-like artificial intelligence, and then we amplify it by 1000x following Moore's law for a couple of decades. Kill switches would become useless. We'd just have to teach them to value life more than anything, before it's too late.
(https://en.wikipedia.org/wiki/Three_Laws_of_Robotics#.22The_...)
Also see this discussion: https://en.wikipedia.org/wiki/Ethics_of_artificial_intellige...
Does anyone know how that law is flawed? Unless I'm remembering it wrong.
A robot must obey the orders given to it by human beings, except where such orders would conflict with the First Law.
1) Plants survive on their own and in groups. Therefore, plants have the right to live. (Don't step on the grass!)
2) Bacteria survive on their own and in groups. Therefore, bacteria have the right to live and should not be eradicated. (Sorry, hand sanitizer.)
3) Cows survive on their own and in groups. Therefore, cows have the right to live and should not be slaughtered. (So much for that steak last night.)
So yes, under this thesis, AI & robots have a right to live, regardless of their threat or benefit to society and humans.
I don't think so. That my friend, is the only thing that separates us from pure animals (i.e., morality).
(I do not wish to open the can of worms of consciousness, soul, etc., and etc., so, HN, please refrain.)
i mean if some machine is not usefull for nothing and it wastes resources than you have right to shut it down if you can, no matter what IQ it has.
When the majority of the human population understands technology and its ever-changing limits, the fear and stigma towards artificial intelligence comparable to our own will subside and only then will sentient robots be accepted.
It's human nature to fear the unknown. It's gotten us to where we are today. But I think there comes a point in our timeline where we overcome all fear with our ability to understand and predict. Ironically enough, this will probably require the advancement of (and potentially cooperation with) A.I.
http://lifeartificial.com/
I think a far more interesting question is: will a human ever be put to death for "killing" an artificial intelligence?
The first step towards becoming more emotional will be naming them. I remember reading a couple of years ago about that cleaning robot Roomba, and how people started naming them, and when they were giving them away for repairing, they demanded the same robot back, not a replacement.
http://www.msnbc.msn.com/id/21102202/ns/technology_and_scien...
http://gizmodo.com/5483750/peoples-emotional-attachment-to-r...
A few years ago I was reading sort of a joke, that by 2050, Massachusetts will be the first state that will allow human-to-robot marriages.
There's already this:
http://www.youtube.com/watch?v=9q4qwLknKag
If some countries that have capital punishment don't get civilized by the time we have human-level AI, then yes, that would be logical. There is no difference between a mind run on a biological brain and a silicon brain. Killing either is murder.
A truly intelligent and self-aware computer will be created by simply simulating a human brain on a powerful computer. I can't see how that mind would have any less rights than its biological counterparts.
So the answer is: Yes, it'll have the exact same rights, as we do. Morally, that is, because the law might take time to catch up.
If the copy is allowed to advance one nanosecond and is then deleted, leaving the original running, has a murder been committed? What if the copy never runs at all? What if the original is deleted instead of the copy?
What if the copy receives identical input, and thus has identical state as the original, up to the point it is deleted? What if some aspects of the copy are "optimized" by reusing the results of the original? What if the entire copy just mirrors the state of the original, without recomputing anything? How can we say exactly how many copies actually exist?
What if the original simulation modifies the copy so that it wants to kill itself, which it does, once it starts running? What if it's modified to just not care whether it lives or dies? What if the copy is modified in various other ways? Are some kinds of modifications ethical, while others are not?
Point being, this course of AI development does not spare us having to completely rethink our models of ethics.
1) yes
2a) yes
2b) It never lived, it was a "frozen corpse", and since you can make copies at will, just having a copy doesn't constitute a new life until it's run.
2c) If the original has been deleted at the same state as the copy is started, i.e. there's no interruption from the subjective point of view of the copy, then it's just a transfer.
3a) Since there's zero information difference, I'd say it is OK.
3b) They're two separate people with equal rights.
3c,d) No computing = no thinking = no life to speak of.
4a,b) The same as if you kidnapped someone from the street and convinced them to kill themselves (say, by drugging them).
4c,d) Very interesting. On the one hand, in case of many modifications, you can't really foresee the consequences, so all should be considered unethical to prevent unintended suffering.
On the other hand - what is human parenting and bringing up if not shaping (or "modifying") a not-yet-sovereign mind. And we have no problem with that. But then, what if some psychotic parent brings up a kid with an implicit conviction that he should kill himself upon reaching adulthood (i.e. legal sovereignty). We'd put them for life in jail. But what if a parent has no bad intentions, but is just a very crappy parent and his kid grows up to be very miserable and in his 20's commits suicide because he can't deal with life. Well, we could pass moral judgment on the parent (if we somehow had the knowledge and thus certainty that it was a result of bad bringing up), but we couldn't really convict them in a court of law.
So yeah, we're having a serious split here between moral judgment and what's practical. (Much like pro- and anti- abortion.)
So I'd say playing with the brains of sentient beings is unethical in general. Practically though, there's no way to stop a "virtual" person from running all kinds of crazy stuff on their private machines. Unless you want to police the internals of all such brain-simulating computers, but then you're conducting mind-control of people, and that's totally unacceptable.
5) Agreed, but I think if we engage in it already, we can get to some very interesting conclusions that we might find useful even in today's world.
How long does it have to run before it's alive? If the interval is short enough then there will be no change in state whatsoever. So in what way does the mind state have to change in order for it to have lived? What if that change is so simple that you can calculate it in your head? Have you created a new life just by thinking about what will happen next in the simulation? By writing it down?
3a) Since there's zero information difference, I'd say it is OK.
So you would have no objection to being killed if there was an identical copy of you in existence somewhere? Why does that matter? Why do you care more about this copy of you than you do about anybody else? Why should a stranger care that there is at least one copy of you?
4c,d) Very interesting. On the one hand, in case of many modifications, you can't really foresee the consequences, so all should be considered unethical to prevent unintended suffering.
But you are modifying your own mind all the time, by forming memories and learning skills. To some degree, you choose what you think and experience. Imagine that you have ultimate control over every facet of your mind. You live in a computer, so there is no "natural" mental life. You must choose exactly how your mind will progress.
If you make a frozen copy of yourself, it hasn't diverged yet, so you can modify it however you like, since it's you.
And if new, unique people are to be created at all, someone will have to choose how their mind works. There is no longer a "natural" reproductive process to decide this. How can choices be made in the new person's interests when it is those very interests that are being chosen?
3a) By killing me and starting an identical copy of me somewhere else, you're creating a discontinuity that infringes upon my identity (or same-ness) because there would be a very strong discontinuity in inputs simply because of the different location.
Moreover, a third party exercising control over any of my copies (and especially the one that is running) is in violation of my sovereignty. [That's also why a stranger should care about my copies.]
4c,d) That's precisely it: I am doing it! I can do with myself whatever I want. And yes, I do imagine having the control about every facet of my mind. I actually dream about it. :)
Now, when I manipulate a copy of me, that's a little different, because as I've stated earlier, a copy spun off to run independently is sovereign from its "original". That's why manipulating own copy should morally be considered equivalent to manipulating a third party, assuming that copy will be let run separately.
To your last point, I agree, that's the most delicate part - creating new beings. As far as copies are concerned, my immediate instinct is that only identical copies should be considered ethical, since any modifications can have unpredictable consequences. But I'm not sure this should be enforced by the AI community / data center, because I value personal sovereignty and habeas corpus virtualis above all else. ;)
I won't speak of creating completely new, original beings, because that topic deserves a ton of thought and a philosophical treatise of its own. But theoretically, you could try to mimic natural processes of child brain development. Of course, as you've pointed out, someone has to choose the parameters of that developmental algorithm, just as a search engine is never really objective, because people wrote it and made some choices in the process. But that last example also shows we can have some trust in such choices for practical reasons.
On a side note: I would highly recommend the book "Diaspora" by Greg Egan. The first chapter (Orphanogenesis) speaks about the development of a virtual mind and to me is among the best works of science fiction of all time.
One of about a dozen questions on there I've marked as "important to me" is one on AI rights. And interestingly enough, I'd say the site's suggestions have become markedly better since I did that; I think that question is a barometer for a lot of things about one's morals and interests.
(Granted, anyone trying to make an AI will aim towards something like us, but considering how we still haven't developed AI yet, I imagine the first successful one will be more of a lucky accidentally-on-purpose than anything strictly-designed; even with humans making it, the first success will be quite non-human.)
Of course, this brings up the troubling thought experiment: if you could copy your own brain, and install it in a new body after your death, would that still be you? Would that satisfy your "right to life"? Our present theory of consciousness is insufficient to answer this question conclusively. My opinion is that consciousness is epiphenomenal.
http://www.amazon.com/gp/product/B002HR17ZG/ref=as_li_ss_tl?...
1) No rights. We can do whatever we want, just like they are our property and we can treat them as pure objects
2) Some rights. Think of it like animal rights, but we still own them
3) "Official" human-like rights. They are free, but there would be "racism" against them at first
4) The exact same rights as a human, officially and unofficially.
Then things might start to change again when they start becoming drastically more intelligent than us. Then it becomes much harder to predict. At first glance, we could say it will be bad for us, but I choose to believe the optimistic side that the smarter they will be the more tolerant they will be, too.