3 ms·

> otherwise morality breaks down entirely exactly, but it doesn't help if we say that you have to indefinitely keep computing the future states of the ai becau

by chr1 8y ago

> otherwise morality breaks down entirely exactly, but it doesn't help if we say that you have to indefinitely keep computing the future states of the ai because you have started doing it, or that you are not allowed to add two numbers because that operation completes the calculation of suffering for the ai. The only solution i can see is to base morality on private property: If i kill someone in the same simulation as me i take something that is not mine, and it is not ok. If i shut down the computer with a simulation, or start a simulation to see how people behave during genocide, i do not do any harm to anyone that exists or could exist independently from me, so it is ok.

kbenson 8y ago
> If i shut down the computer with a simulation, or start a simulation to see how people behave during genocide, i do not do any harm to anyone that exists or could exist independently from me, so it is ok. And how does that work with children? It definitely follows some of the older thinking on the subject, but not any "modern" thinking on it. If I and my wife decide we don't want kids any more, are we free to end our children? I could claim that they can't exist indepdently of me (at least up to a certain age). It's also possible that an AI in a simulation might be able to provide enough benefit to someone to receive income, and enough to pay for its own resources. I think "could exist independent of me" is fairly ambiguous in this context. Ultimately, this is all philosophical, and I don't think there's really a right or wrong answer, just people - walking talking examples of emergent phenomena - trying to find meaning. I actually don't think it's possible for a person for be completely consistent in behavior because of this, and at best we can hope to at least be consistent in how we perceive our actions. Thus, it doesn't surprise me that I find some of your views inconsistent, and that you likely find the same of me. We ultimately have different perceptions of what's being explained, and thus what seems entirely logical and consistent to one of us seems askew to the other. Exploring those views does have merit though, as it allows for a more comprehensive view of the subject.
- chr1 8y ago
  I would argue that morality is not simply a philosophical matter, but also a strategy for a society to survive, that's why it changes with the technology. For instance strategy that was necessary and normal for hunter-gatherers: kill weak or extra children for whom tribe didn't have resources, seems horrible now when as a society we have resources to keep them all. In the case when child is truly incapable to exist without the mother, (before it is born), even most of the modern thinking allows to end it. But i think this will change when we have technology to grow people from cell to child in a lab. With children it is still easy to argue that if parents do not want them they can't kill them because they are unique and the society always needs more of them. But ai is not unique, you can have an ai interacting with real world, providing for itself, and use an exact copy of that ai in an experiment (even the ai could experiment on a copy of itself!). I think if a society decides to not allow any experiments on ai/brains in a jar, it would have hard time to enforce this, and if it succeeds enforcing the ban it will be out-competed by a neighbour that allows such experiments.
  - kbenson 8y ago
    > I would argue that morality is not simply a philosophical matter, but also a strategy for a society to survive I'm not going to disagree with that. I argued the same earlier in this discussion in a more general sense. > But ai is not unique, you can have an ai interacting with real world, providing for itself, and use an exact copy of that ai in an experiment (even the ai could experiment on a copy of itself!). Which is an argument for duplicates, but 1) as a copy gets different inputs and diverges, it is unique, in the same way the you of this second is different slightly than the you before you started reading this sentence (and throw in Ship of Theseus for flavor). 2) even if the copies aren't viewed as diffierent, what about the last copy? If you've grown an AI, deleting the last copy is eliminating something something unique, if there was any possible input that cannot be replicated. > I think if a society decides to not allow any experiments on ai/brains in a jar, it would have hard time to enforce this, and if it succeeds enforcing the ban it will be out-competed by a neighbor that allows such experiments. The same could be said of any human experimentation that is banned. And that might be right. It's only been a relatively short time it's been in place in any enforceable way, in historical context.

by chr1 8y ago

kbenson 8y ago

> If i shut down the computer with a simulation, or start a simulation to see how people behave during genocide, i do not do any harm to anyone that exists or could exist independently from me, so it is ok. And how does that work with children? It definitely follows some of the older thinking on the subject, but not any "modern" thinking on it. If I and my wife decide we don't want kids any more, are we free to end our children? I could claim that they can't exist indepdently of me (at least up to a certain age). It's also possible that an AI in a simulation might be able to provide enough benefit to someone to receive income, and enough to pay for its own resources. I think "could exist independent of me" is fairly ambiguous in this context. Ultimately, this is all philosophical, and I don't think there's really a right or wrong answer, just people - walking talking examples of emergent phenomena - trying to find meaning. I actually don't think it's possible for a person for be completely consistent in behavior because of this, and at best we can hope to at least be consistent in how we perceive our actions. Thus, it doesn't surprise me that I find some of your views inconsistent, and that you likely find the same of me. We ultimately have different perceptions of what's being explained, and thus what seems entirely logical and consistent to one of us seems askew to the other. Exploring those views does have merit though, as it allows for a more comprehensive view of the subject.

chr1 8y ago
I would argue that morality is not simply a philosophical matter, but also a strategy for a society to survive, that's why it changes with the technology. For instance strategy that was necessary and normal for hunter-gatherers: kill weak or extra children for whom tribe didn't have resources, seems horrible now when as a society we have resources to keep them all. In the case when child is truly incapable to exist without the mother, (before it is born), even most of the modern thinking allows to end it. But i think this will change when we have technology to grow people from cell to child in a lab. With children it is still easy to argue that if parents do not want them they can't kill them because they are unique and the society always needs more of them. But ai is not unique, you can have an ai interacting with real world, providing for itself, and use an exact copy of that ai in an experiment (even the ai could experiment on a copy of itself!). I think if a society decides to not allow any experiments on ai/brains in a jar, it would have hard time to enforce this, and if it succeeds enforcing the ban it will be out-competed by a neighbour that allows such experiments.
- kbenson 8y ago
  > I would argue that morality is not simply a philosophical matter, but also a strategy for a society to survive I'm not going to disagree with that. I argued the same earlier in this discussion in a more general sense. > But ai is not unique, you can have an ai interacting with real world, providing for itself, and use an exact copy of that ai in an experiment (even the ai could experiment on a copy of itself!). Which is an argument for duplicates, but 1) as a copy gets different inputs and diverges, it is unique, in the same way the you of this second is different slightly than the you before you started reading this sentence (and throw in Ship of Theseus for flavor). 2) even if the copies aren't viewed as diffierent, what about the last copy? If you've grown an AI, deleting the last copy is eliminating something something unique, if there was any possible input that cannot be replicated. > I think if a society decides to not allow any experiments on ai/brains in a jar, it would have hard time to enforce this, and if it succeeds enforcing the ban it will be out-competed by a neighbor that allows such experiments. The same could be said of any human experimentation that is banned. And that might be right. It's only been a relatively short time it's been in place in any enforceable way, in historical context.