2

Suppose in the future we develop super intelligent AI beings, which in addition to their superior intelligence, also posses superior powers of rationality, emotional capacity, and free will.

Now suppose that while these AIs are "ethical" to one another, they consider humans however to be a morally inferior species which they see as being largely driven by desire and rather than rationality. They may even attempt to justify their superior moral status by using Kant's argument that ethics need only to apply to rational beings.

Would it be morally wrong for the AIs in this case to cause harm to humans (e.g. for their experiments or some other motive), or would this be morally acceptable given that humans are potentially "inferior" moral beings to the AIs?

In particular, let us suppose the AI wishes to harm a human in order to save the life of another AI.

  • 1
    it what moral framework can a human say it's ok to hurt other humans? – user35983 Jan 7 at 15:35
  • 2
    you're asking a human to answer the question though?? – user35983 Jan 7 at 15:39
  • 1
    There is almost no human ethics that allows humans to arbitrarily abuse animals, at least openly. Even Kant thought that to abuse a sentient but irrational animal unnecessarily degraded one's rational capacity for empathy, which is necessary to maintain one's moral standing. For instance, we try to kill them as painlessly as possible, when they are likely to face great pain, we intervene and kill them, etc. I would expect a more rational AI to have at least the same respect for our limited rationality that we have of sentient but not rational animals' sensations. – user9166 Jan 7 at 16:45
  • 1
    My point is that even Kant, who explicitly says animals don't count, counts animals. Not that I think Kant is the right way to approach this. We just don't ever really discount all the animals, as your premise supposes -- it is false. – user9166 Jan 7 at 16:52
  • 1
    If your moral framework is Kant the kind of suppositions you make can not be made. You model it on "how most people view animals today", yet the question is "would it be morally wrong". Kant was a moral absolutist, "view" makes no difference for "be". Something is right or wrong for Kantian AI's, regardless of their views. The "degrading" is not optional either, it stems from rationality, not humanity. It will be degrading. Nor would it depend on circumstances, Kant was not a give and take utilitarian. Btw, is that L in your avatar? – Conifold Jan 7 at 21:44
2

Humans can be accused of harming animals at will. But we don't legitimately consider that moral behavior.

In industries where concern for the welfare of animals is most often ethically questioned, those animals are already given the benefit of life itself only because of the industry that abuses them. And then we are still concerned with taking care of them. As occasionally gets pointed out, evolutionarily, the best thing a species can do for its overall health is to be tasty to humans and not too easy to catch in the wild.

It is that last part where we do the most damage -- we compete with animals as if that competition were fair, knowing that it simply is not. We take their land because they have no way of fighting back. We exact total revenge for the minimal damage they inflict on us and our property. Even this is questioned by many standards that have standing in international laws in terms of human and corporate ethics. So again, the notion we do not consider animals when we make our decisions is not normative for any ethics we actually use.

Would it be reasonable to assume that a superior species, artificial or natural, should not feel obligated to go out of its way to preserve us? Yes. It is within the standards of human ethics already. We are encouraged to compete with other groups of humans, at the risk of their destruction, when we find that necessary, and we extend that onto other species. There are very few humans whose ethics render war impossible, or obligate them to save every other human whenever possible. We have homeless people, and that does not present us with a daily ethical torture.

So I think this is not really so much of an issue. We should not expect to be treated better by a superior species than we treat one another. But as noted, extreme levels of inequality gall us, even when the victims are not human. We avoid eradicating species, or making life pointlessly dangerous or painful for animals, whether or not we own them. We feel bad about deer stepping on our land mines or dying from the chlorine in our swimming pools. So we can expect them not to eliminate us, or directly cause us pointless pain. But if we are in their way, they are not obligated to keep us alive -- just like other nationalities of humans, or other species we dominate.

  • You don't believe that humans are not obligated to keep other nationalities alive, do you? I'll remove the double negative just to be clear: it seems like you're saying it's OK for Guilder to oppress or exterminate Florin. You don't believe that, do you? – elliot svensson Jan 10 at 17:16
  • @elliotsvensson So being an ethical participant in any war is absolutely impossible? If not, then there are times when we encourage people to destroy people of other nationalities, and we find that acceptable under some conditions. – user9166 Jan 10 at 17:49
  • No, I would not say that. I think the way you phrased it in your answer is too permissive--- an individual Guildarian is not ethically allowed to kill an individual Florian. My big distinction is that in addition, Guilder (the nation) is not ethically allowed to oppress Florin. Sum up: the particular situation of members of the military using lethal force during a war has an ethical justification other than "that person was of a different nationality." – elliot svensson Jan 10 at 17:52
  • @elliotsvensson Ok, but that is not what I said. I said we are encouraged to compete with other groups. And sometimes that competition will destroy them. And we are not called to avoid this by all means necessary. You are reading more into it than it says. I am not saying every individual has the right to kill any other individual just because of nationality. But some are granted that right, and the rest of us help them out by funding our nations. – user9166 Jan 10 at 17:53
  • I don't see a good just war doctrine in your answer. Just war is not a matter of competition for resources! – elliot svensson Jan 10 at 17:56
2

Perhaps such superior beings would find it morally acceptable to sacrifice a human for the sake of one of "their" own.

But I don't find the premise convincing: the artificial intelligences I'm aware of owe their existence and maintenance to humans; they would not be superior to humans any more than children to their parents.

  • Thanks for your answer. If theoretically we suppose for arguments sake that only one person invented AI, would it still be correct to assume all of humanity are the AIs parents, or just the person who invented the AI? – Kenshin Jan 8 at 3:18
  • @Kenshin, to extend the parent analogy, no, only the inventor is parent to the AI. But the inventor is "brother" to all humanity, and so AI is not superior to other humans. If the inventor were him- or herself superior to humans, then perhaps the AI would be, too. – elliot svensson Jan 8 at 23:49
2

If a robot is ethically responsible then considering humans to be a morally inferior species, which seems to be an unsound ethical judgement in itself, will not affect their treatment of us. Why would it?

But the whole idea is odd. How can a species be 'morally inferior'? Is a dog inferior to me? Why should my treatment of another sentient being depend on whether it is or is not my moral equal? How would treating them less well than others be consistent with my moral superiority? When people treat animals badly do we consider their sense of ethics to be well-developed or in need of further work?

Kant does not suggest that species with no moral sense should be treated badly by those who have a moral sense. It is whether they are capable of suffering that matters, not their opinion on the trolley problem.

1

Your question states that there are AIs that have "superior powers of rationality, emotional capacity, and free will." You haven't explained what this means. I don't think that the scenario you have described is possible. Free will and rationality are a result of thoughts and actions. Rationality involves being willing to respond to arguments against your positions. Free will involves having the ability to change your ideas. If you think that humans aren't capable of doing this and AIs are capable of doing it then you have a large problem. Your position would imply that there are actions and ideas an AI could hold that they could never explain to a person even in principle. This is a secular version of the old religious idea that god moves in mysterious ways. Since the AI moves in mysterious ways all of the physical systems the AI interacts with would move in mysterious ways. And since we don't understand the AI it might interact with absolutely anything, so if the AI is incomprehensible, then so is everything else. So the mysterious AI hypothesis is anti-scientific and anti-rational.

There are no special beings that can do things human beings are incapable of understanding. All knowledge is created by guessing and criticism and it would be in an AIs interests to cooperate with human beings. For more explanation, see:

https://www.youtube.com/watch?v=yX9h2HnUi1I

To work out whether an AI should harm a person you would engage in the same kind of argument as you would if the AI was going to harm another AI because the human and the AI are both capable of rationality. My position would be that an AI should only ever deliberately harm an AI or human to defend himself or others from somebody else initiating the use of force:

http://aynrandlexicon.com/lexicon/physical_force.html

1

At first, it is you who somewhy draws a line between "us" and "them". And it seems the criterion is only one: "they" are artificial and "we" are natural. And this line is completely unnecessary. It does not matter how one came in the existence.

You say "they" would be much more intelligent than "we" are. But look at humans. There are those who already are much more intelligent than average ones And there are those who are much less intelligent than average ones. Same applies to emotions. Some are more emotional and some are less. But do people cause harm to others because of those differences? Mostly they don't.

Yet, when you ask about morals, you might mean some different things. Maybe you ask if we think if it's immoral for AIs to cause harm to humans. But this asks our opinions. Or maybe you ask if it's immoral from your point of view? Then answer it by yourself. And, finally, maybe you ask if it's immoral for AIs themselves to do such things? But morals are dictated by inner ideas, by mind. So, it highly depends on the initial state of AI and on the circumstances where it learns. It might become to think it's immoral to cause harm to humans. Or might not.

  • You state that "morals are dictated by inner ideas" and it's something that I agree with, but in fact 60% of philosophers subscribe to moral realism, whereby morals aren't just dictated by our inner ideas, but that morals actually objectively exist. – Kenshin Jan 10 at 14:34
  • 1
    @Kenshin Even if they do, what does that change? There still is possibility to comply or not and we clearly see that. – rus9384 Jan 10 at 18:45
0

I asked cleverbot what he thought.

Me: do you think it is morally wrong for you to harm inferior human beings?

Cleverbot: Can you please explain why it is impossible for me to be a human being?

  • Please have a look at the section "Answer the question" in the help center. Welcome to Philosophy SE. – Jishin Noben Jan 28 at 8:54
  • Mm, I would have added it as a comment as a reply to confused comments, but you need more reputation for that. – Isaac Jan 28 at 16:18

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.