Non-conscious on conscious violence

Abstract

A common objection against autonomous weapons systems is that their use violates human dignity. Machines, it is said, should not decide whether a human being lives or dies. If they do, our dignity is violated. But why? This essay frames the problem as one of “non-conscious on conscious violence”. Criticisms of autonomous weapons should not rely on the fact that they do not think, but on the fact that they are not conscious. I then take a closer look at arguments for why being killed by non-conscious beings violates human dignity. I show that the most prominent arguments fail to establish that it necessarily does. I argue instead that the debate on AWS is a debate on how we can express our respect for conscious life. This we must settle before the dignity-argument can be advanced seriously.

1. Introduction

Technology has increasingly removed humans from the violence they inflict. Over centuries, we went from killing others with sticks and stones to guns and intercontinental ballistic missiles. Today, we may never see the person we harm – they may not even be in the same country as us.

Soon, autonomous weapons systems (AWS, sometimes also called LAWS for lethal autonomous weapons systems) may further remove humans from the violence they cause. These systems could soon be deployed on battlefields or inside countries – for example to assist law enforcement. What makes them novel is that they make their own choices. These weapons do not simply follow their creator’s will or that of the soldiers who deploy them, but could, through sophisticated machine-learning, develop their own strategies on battlefields and beyond. Since these strategies would sometimes require killing someone, this would create an environment where “a collection of microchips and software will decide whether a human being lives or dies.” (Guizzo and Ackermann 2016, p. 40).

Naturally, this development draws criticism. Can AWS accurately distinguish between combatants and civilians? Will they create a global arms race? Could they make wars easier to launch and thus more frequent? These are important questions, but not the topic of this essay.¹ Instead, I focus on one particular criticism of AWS, namely a principled objection against letting machines decide when to kill humans. According to this position, even if AWS could correctly identify targets, would not create an arms race or make wars more likely, they should still be rejected, because no matter the circumstances, machines should not have the power to end human life. Granting them this power, the argument goes, would violate the dignity of the humans targeted by these machines (Heyns 2013; 2017, Amoroso and Tamburrini 2018, Sharkey 2019). I will call this the “dignity-argument” from now on.

The dignity-argument has different types, some highlighting that “death by robot” reduces people to data-points, others claiming that it takes away our ability to plead with the enemy. The intuition that “death by robot” is wrong has some force and the goal of this essay is to analyze it closely. To be clear, I do not provide a comprehensive ethical analysis of AWS. I closely examine the dignity argument and thus, even if I show that it fails, there may be other successful arguments against AWS.² Nonetheless, this essay aims to advance the debate on AWS by uncovering the dignity-argument’s underlying points.

I focus on the dignity-argument for two reasons. First, it is one of the strongest objection against AWS, since it would hold even if they had positive consequences (Rosert and Sauer 2019). If technological progress makes these weapons more sophisticated, arguments about, say, successful targeting may wither away. The dignity-argument would remain. Second, I believe that the argument rests on implicit points, and illuminating them would improve the debate surrounding AWS.

I believe that it is an argument about consciousness. Fundamentally, the problem of AWS is a problem of “non-conscious on conscious violence”. What strikes us abhorrent, I propose, is the idea that non-conscious beings can decide to kill conscious ones. The reason this feels wrong, I suggest, is the absence of phenomenal consciousness. Contrary to what critics sometimes proclaim, machines may be perfectly capable of valuing and respecting human life. However, they cannot value it the way a conscious human being can.

To test the dignity-argument, we must answer two questions. First, what grants human beings dignity? Second, which property of robots violates it when they kill a human? I argue that both parts relate to consciousness how it grounds moral status. My analysis elucidates that when we object to AWS with the dignity-argument, we object to “non-conscious on conscious violence”.

I will outline the connection between consciousness and the AWS debate in the next two sections. Then, I ask if the dignity-argument can withstand critical scrutiny. Do machines automatically violate human dignity if they kill us? And if so, why? I argue that the most promising approach is that when we launch non-conscious agents to inflict violence on conscious ones, we undermine norms that are important to us. However, I argue that this does not show that non-conscious on conscious violence necessarily violates human dignity.

In the following pages, I mostly rely on the literature on AWS in war. However, since the deeper problem is non-conscious on conscious violence, the conclusions of this essay apply to every domain where non-conscious systems gain the power to kill conscious beings.

2. Consciousness and Human Dignity

Autonomous weapons contain the potential to revolutionize warfare³ by deciding when to end a human life. Naturally, such a step causes ethical debates. War and the killing of humans are already problematic, and automated weapons further complicate things. As Heyns (2017, pp. 47f.) put it:

“As human beings, we have grappled since time immemorial with making sense of death at the hands of other humans or as a result of nature. We are now confronted with a new case: death by algorithm. This challenges many of the assumptions engrained in the ethical, legal, religious and other normative systems of the world.”

The dignity-argument asserts that in situations where we have to kill humans, we owe it to them that they are not killed by a machine – more precisely, that the machine does not make the decision to kill them. If they do, then the dignity of the person killed is violated.⁴ This line of thinking has two parts. First, humans have something worth protecting (dignity) and second, robots deciding to kill us violates it. This section deals with the first part. Put simply, what is dignity and why do humans have it?

Dignity is a contested concept. Not only do different cultures stress different conceptions – an Islamic approach will differ from a Western one – but the idea of human dignity also changed over time. It can also have varying interpretations within one society (Sharkey 2019). Still, we usually assume that humans have it. It is, in other words, part of a human’s moral status. Moral status means that those who have it are morally considerable – we have moral obligations towards them. We have obligations towards humans, but not towards rocks. Two important points attach to moral status. First, it primarily prohibits us from treating people in a certain way. That is, if an entity has moral status, we are forbidden from treating it any way we might want. Second, we are constrained this way for the entity’s own sake. I may not treat a painting some ways, but not for the painting’s sake. I do so for the people who value it. By contrast, we refrain from treating humans a certain way for their own sake, not because of other people who value them (Shepherd 2018, pp. 14f.).

Like moral status in general, dignity protects us from certain treatments. Heyns (2017, p. 62) argues that “it is dignity that gives life its meaning”, and that it “emphasises the notion that each person is entitled to be treated according to his or her own full merits”. Finally, dignity requires that those who have it are “taken seriously as separate and irreplaceable individuals”. In a similar way, Habermas (2010) outlines that dignity prohibits us from treating human beings in certain ways. Dignity was a reason that the German Federal Constitutional Court decided the armed forces could not shoot down a passenger aircraft that had been captured by terrorists. Even if the airplane, now turned into a weapon, threatened people on the ground, respect for dignity meant that the state could not use the people on board as a means to protect others by shooting them down.

Dignity and moral status both prohibit certain ways of acting towards others. To be clear, dignity and moral status are not the same. Some beings may have moral status but no dignity – although having dignity means having moral status. It is assumed that humans have dignity, but one may ask if non-human animals have it too – unfortunately, this is a question for another essay.

In any case, the above only tells us what dignity requires of us when we interact with those who possess it. However, we still do not know where it comes from. It may be clear humans we have it and (somewhat) clear what it is, but why do humans have it? One candidate for the source of dignity – and the one I defend here – is consciousness. While the connection between the two (or rather between consciousness and moral status) is still debated (see Levy 2014; 2024, Shepherd 2018, Lee 2024), philosophers usually accept that consciousness is relevant for having moral status (Cutter 2017).⁵ What consciousness actually is remains equally difficult to pin down. I will not use a complete theory of consciousness, simply because nobody has succeeded in crafting it. But as Mosakas (2021) argues, one can discuss the ethical implications of consciousness without one. What I will use in this essay is consciousness as a “what-it-is-like” quality (as outlined by Nagel 1974). This refers to phenomenal consciousness, which is the ability to have subjective experiences with a particular phenomenal feel to them. There is “something-it-is-like” for me to be in pain or be joyous, to eat a delicious meal or to hear a piece of music that makes me nostalgic about my teenage years. As many philosophers agree, these qualitative states matter morally. Because I can have them, I am a moral patient (Levy and Savulescu 2009).

I take this rather minimal approach to advance my analysis. Of course, the debate on consciousness and moral status is vast and there are many fine-grained distinctions. But most of them do not concern us here. For instance, one may ask if the intensity of phenomenal consciousness determines the strength of obligations we have towards conscious beings (see Shepherd 2023, Levy 2024). One may also wonder how we should treat beings that are identical to us but have no phenomenal experiences (see Shepherd 2024).⁶ However, I am focused on human beings and on the way we ought to treat them. I am not making comparisons with other conscious entities.

What interests me here is the connection between consciousness and dignity – more precisely how it grounds dignity. An answer was provided by Kriegel (2017; 2019). Kriegel argues that because of consciousness, we have dignity. Another way to phrase it is that consciousness grants us “recognition respect”, which is respect we receive because we are persons – not because of, say, achievements. This kind of respect is similar to dignity. Like dignity, it applies to in all persons alike. Recognition-respect also prohibits certain ways of treating those who possess it (Kriegel 2017).⁷

Why would consciousness be the grounds of dignity – or at least a necessary component? For Kriegel the reason lies in how we perceive other conscious beings. While we may sometimes have negative attitudes towards them – such as towards a waiter in a cafe – reflection about these attitudes should make us realize that they are not appropriate. The waiter is not a legitimate target of callous or belittling attitudes. They are not merely an object in our own life, not only a background-character in our story. They are their own story. Just like us, they have a complex, subjective view on their own life. To paraphrase a metaphor of Kriegel, they are not merely a planet that orbits us, but their own sun (Kriegel 2017).

To explain the attitude we should take towards others, Kriegel uses the “norm of incorrigibility”. That is, because we have privileged access to our own mental states, others should not dare to correct us about them. If we abide by this norm, we must necessarily assume that our counterpart has a certain dignity we cannot violate (Kriegel 2017).

This explains why consciousness and dignity are interlinked. The stance that other people are their own main characters (for lack of a better description) should, if we carefully test our intuitions, make us hesitate before we treat them in some ways they may not appreciate. And since we should not correct them about their own mental states, what they appreciate or is their decision, not ours.⁸

Kriegel’s account is a start – and a good one – however, one issue is that we cannot accurately define, at least for now, if dignity includes the right not to be killed by a robot. In any case, it is difficult to find a comprehensive source of dignity, given the contested nature of the topic. Still, it seems intuitive that consciousness – and the complexity of human consciousness – is necessary for dignity. It is also a position that critics of AWS should be interested in defending. After all, they would not argue that dignity comes from being an end-setting entity (the way Kant would) nor that it is something that adheres to humans simply because they are human. The problem with the first approach would be, as Kriegel (2017) points out, that complex, non-conscious machines can set ends and would thus possess dignity. By contrast, conscious beings who cannot set their own ends – such as immobile beings who have no ability and no desire to affect their environment – would not have it. This seems like an unacceptable conclusion. The problem with the second approach is that we would bar other conscious life forms from having dignity. We may defend this with non-human animals, although this is far from guaranteed. However, a critic of AWS would likely agree that if we discovered other lifeforms with advanced consciousness that matched humans, we would not be allowed to wage war against them with AWS.⁹ Thus, it is fruitful for the dignity-argument to rely on consciousness.

However, we still do not know what dignity is nor why death by algorithm violates it. This is the second part of the problem: the properties of the agent killing us – the robot. While we acknowledge that killing conscious beings is wrong, in some situations it may be justified. That situation may appear in war or it may arise from the duty to protect the life of others – such as by police shooting a terrorist who is about to storm an elementary school. We are morally allowed, even obligated, to end conscious life in some situations. As critics note, however, we must follow certain procedures – one being that the decision to kill is taken by conscious agents.¹⁰

I believe that if the moral status of humans stems from consciousness, the fact that robots should not kill us stems from the fact that they are not. This creates the question that I will examine in the rest of this essay: what exactly is it that violates our dignity when a non-conscious agents makes the decision to kill us?

3. Non-Conscious Machines and Violations of Dignity

That AWS violate dignity is often repeated (see Heyns 2013; 2017, Sharkey 2019). The previous section has argued that as beings with complex consciousness, humans have dignity. One component of dignity is that the decision to kill us must be taken by conscious beings. Since machines are not conscious, they cannot be allowed to take our life. But why should this be valid?

It surely is intuitive, at least according to a UN report which identified “an instinctual revulsion against the idea of machines ‘deciding’ to kill humans.” and argued that this revulsion “might be considered a reaction to a potential violation of human dignity” (UNIDIR 2015, p. 7). That the machine decides to kill, and does not merely implement human decisions, means that analogies with past instances of mechanized killing are not applicable. Humans have developed various methods to kill many people with the help of machines, nonetheless, the decision was always taken by them. While technology may diffuse responsibility between actors, humans are making the decision. Automated weapons are different. While humans deploy them, they cannot foresee how the machine will behave in complex situations. A decision to kill, not foreseen by any human operator, may be taken by the machine. We may always assign some responsibility to commanders them or manufacturers, but we cannot deny that part of the causal responsibility for the death lies with the autonomous system.¹¹

A preliminary question is who’s dignity is violated – with the most straightforward answer being the dignity of the people killed.¹² But why would they care how they die? They may care whether their death is long and painful or not, but would they care if they are killed by a robot or a human? A possible answer is that people maintain dignity even after their death – which is why we are strongly against desecrating corpses. We can thus violate someone’s dignity without them knowing how they died. In a similar vein, one may violate their partner’s dignity by cheating on them – even if they never find out.

However, we might also say that AWS violate the dignity of a person before their death, because they make their death “arbitrary”. As (Heyns 2017) argues,¹³ human dignity protects us against “arbitrary death” and AWS cannot be but arbitrary. To not be an arbitrary kill, the decision behind it must be taken by humans. Otherwise, the death becomes the result of algorithmic decision-making. Another way to frame this is that we do not have the right to kill others without applying our minds – and thus, we cannot delegate this right to machines who kill without applying theirs. Only humans can fully appreciate life’s value, while a machine could not respect the value of life or comprehend the significance of ending it (Amoroso and Tamburrini 2018, Sharkey 2019). The point is succinctly put by Renic and Schwarz (2023, pp. 335f.):

“The human dignity challenge of LAWS stems not from the material character of the targeting effect, but rather from the moral character of the targeting process. Humans deserve to remain free of predation from systems that lack the capacity to properly assess and weigh their moral worth. LAWS fail to meet this standard by virtue of their inability to recognize and act upon the full range of factors that render an individual morally liable—or not—to lethal targeting. […] Computational objectification of the sort LAWS produce necessarily dehumanizes the targeted individual. It reclassifies humans as something less; something that is statistically determined, processed, and rendered actionable. LAWS have no conception, by design, of a subject-object relation, as we humans would, nor do they understand the subjective self in relation to the object upon which they act.”

The problem may not be a lack of mind or unthinking machines, as Heyns points out, but a lack of consciousness. AWS could, if sophisticated enough, go to great lengths to preserve human life. They would, for instance, not be steered by self-preservation and not fire out of fear for their lives. They would also feel no desire to take revenge on enemies. One could imagine a highly sophisticated AWS which does everything to not end human life – even if it incurs great risk of its own destruction. That machine would certainly take the value of human life into consideration – more than an exhausted or stressed soldier might.

Still, this is not enough to make them morally acceptable, as Heyns (2017, p. 60) argues:

“Someone must take the time and apply his or her mind and ‘sign off’ on the conclusion that using force is really necessary as far as possible in this particular case. […] Machines have no understanding of the importance of life, and of the implications of taking it. While some humans also lack these qualities, machines by their very nature can never have it. Cold-blooded – or rather bloodless entities should not have the ability to take decisions on whether to spill human blood.”

Surely, Heyns knows that the problem with autonomous weapons is not lack of blood. It is also not a lack of mind. Current AWS may be “mindless”, in the sense that they can identify and attack targets, but not much else. We may object to such crude examples of AWS. But a more complex machine may have a mind and could set complex goals – the highest being to protect human life at all costs. Such an AWS may kill rarely– only, for example, to protect human beings. This would be a thinking machine that takes human life into consideration. Still, would Heyns agree that death through such a weapon is no longer undignified?

Suppose that he does not, then the most promising way to approach the issue is pointing out that the machine is not conscious. It may “think” about the value of human life – and form complex deliberations about how to preserve it – but it cannot “feel” the weight of its decision. It cannot appreciate life the way a conscious entity like a human can. As critics note, when AWS kill people, the human is reduced to a data point, by a machine that has no conception of what it means to take a life (Rosert and Sauer 2019). The human may be an object that the machine takes great interest in preserving, but they are still an object. To a machine, everyone is. They cannot conceive of people as holders of recognition-respect, as beings with their own complex view on life, the way Kriegel outlined it. They cannot test their intuitions and realize that there are certain things they are not allowed to do to a human being. If we decide that killing is necessary, there needs to be someone who, in some sense, “has to live with their decision”. Thus, Heyns and others do not object to “mindless violence”, but to “non-conscious on conscious violence.” That AWS are “unthinking” is not the problem. It is that they cannot conceive of humans as beings with their own subjective viewpoint – at least they cannot appreciate this in the way we can.

This section has finished re-framing the problem. The principled argument against AWS relies, first and foremost, on the fact that a non-conscious entity kills a conscious one. With this in mind, the next section can take a closer look at common avenues of the dignity-argument.

4. Why Is It Wrong When Machines Kill?

Humans have dignity. Beings with dignity should not be killed by non-conscious beings because they cannot appreciate human life. The question now is: does that argument hold under critical scrutiny? After all, why should we care that the machine does not appreciate life? It does not need this appreciation to protect life – it may do this even better than human soldiers. To accept the dignity-argument, I think we need to explore why this violation exists. After all, it is a truism that non-conscious robots cannot consciously appreciate human life. But if these feelings are absent, why is the dignity of the dead person violated? Why does a dignified death require this appreciation?

One argument by Heyns (2013) states that by deploying AWS, we change our view of other people. The issue is not only that the machine reduces people to data, but that we degrade our fellow humans if we use the machine against them. Heyns worries that we might think of our enemies as vermin that must be eradicated if we can easily kill them with a button that deploys the AWS. His argument, then, is less about the act of killing with machines and more about this power changing how we think about our fellow humans. If we have the power to delegate the act of killing to machines, our enemies may become vermin in our minds. Something that merely bothers us and that we can exterminate quickly and with little hassle – the way we might set up a fly-trap.

However, this relation is speculative. Why should humans, especially those who care for humanitarian law and the value of life, suddenly stop caring about it if they deploy autonomous weapons? The act of killing others becomes much easier, but the decision to launch these weapons might not. Would leaders who are educated with a respect for human rights suddenly speak of people as vermin behind closed doors? It is true that technology can and has changed human behavior (Morrow 2014). And if killing becomes “easier”, it may become more frequent. But this is no principled objection against AWS. Rather, it is a call to ensure that our attitudes towards humans are not affected by this technology. It demand that we instill a strong appreciation for human life in political decision-makers. In fact, AWS could be a wake-up call – in a similar sense in which the horrors of World War II were one. And after all, humans today have the most advanced abilities to kill their enemies. Still, international wars are rarer than before. And while respect for human rights is far from perfect, the situation is surely better than a century ago.

We can take a lot of force out of this argument if we ensure that our attitude towards other humans does not change. We could, for example, ensure deaths by AWS are counted and evaluated. That they are not forgotten and that people who launch AWS must justify their decision. We could work on making AWS more protective of human life. All of this would not only make people safer, it would express of our commitment to the value of life.

A second argument is that those targeted by AWS cannot appeal to their attacker’s humanity and ask to be spared (Heyns 2017, Amoroso and Tamburrini 2018). As outlined in the previous section, the problem is not that the attacker is not human or unthinking. It is again that they are not conscious. However, we can ask why lack of consciousness makes pleading impossible. If a machine were programmed to preserve human life, then it would surely not engage in deadly violence once humans, for example, throw away their arms. Not all AWS may ave this feature, such as intelligent missiles, but many of them may. Those who do not we may ban, but those where this feature is possible may not attract condemnation. In fact, appealing to a robot may be more dignified than to a human. It would, for instance, be more predictable. With humans, surrendering could lead to abuse, torture and death. With AWS, surrendering would lead to predictable and likely benign outcomes. One must also not feel degraded by surrendering to machine, knowing that they are merely performing the required actions that makes the machine refrain from deadly violence.

A third argument is that the dignity of humans who launch the AWS, not those who are targeted, is violated. The reason is that AWS take away their opportunity to be a moral person and to make moral decisions (Sharkey 2019). That seems contentious, since the decision to launch the weapon would still be taken by people who can be moral agents. Today, generals who are far away from the battlefield can be moral agents. Even politicians who are further away can.

These arguments cannot sustain the dignity-argument. In the first case, the problem is with humans and how they react to the power that comes with AWS. It outlines a possibility, not a necessary connection. In the second and third case, the issues that critics outline may not even arise – at least with some autonomous systems.

A fourth argument is that when AWS decide to launch an attack, we express a profound disrespect for the people we target (Nagel 1972, Sparrow 2016). Importantly, this is different from the first argument. We may never consider our enemies vermin or change our attitude towards them in other negative ways.

But is this expression of disrespect certain? We may argue that it is, that using a tool that cannot appreciate life to end it is always a violation of dignity because it signals disrespect. Suppose, for instance, that a newspaper sent a critic to a piano concert, but sends someone who has no appreciation for music. The critic would do her job earnestly and with the best intentions – and yet, she would not appreciate the concert. I think we can say that the newspaper, by sending the critic, has somehow disrespected the pianist. Not only the respect they owe him due to his achievement, but also in the recognition-respect he deserves as a conscious being. We may argue that such things happen, that they are not that serious or that we can apologize. Still, I think it is fair to say that we wronged the pianist.

The dignity-argument appears to be somewhat fundamental then. There is no deeper reason behind it, such as the inability to plead or the way it may change how we think of other humans. It is a violation merely because it signals disrespect. However, we may also ask if this signal is necessary. After all, the signals that emerge from our acts are complex and can be interpreted in various ways. For instance, the pianist above may appreciate that the newspaper sends a critic who cannot seriously appreciate the music but who is widely read, will surely enjoy the concert nonetheless and help the pianist in his career. It depends, in the end, on what aspect of the act we focus. In depends on the story we want to tell.

Just like dignity is a contested concept, there are different stories on how AWS can violate it – or not. What non-conscious on conscious violence asks of us is to devise ways we can implement it into our existing normative framework. If it violates certain norms we hold dear, does that mean that non-conscious on conscious violence is wrong or that we must change the norm?

A final analogy may drive the point home. For decades, liberal nations have launched humanitarian interventions in other countries. These interventions are based on the principle that human beings have equal moral value and a right to protection – even if their own governments disagree (Archibugi 2004; Pape 2012). Putting the practical problems of humanitarian interventions aside, a key issue they have is that some of their practices undermine their own supposed values. Despite rhetorical commitments to universal human rights, Western governments value their own soldiers’ lives higher than those of the civilians they protect. At least they sometimes act like they do. This is especially visible in a prominent tool for humanitarian interventions: air strikes. These protect soldiers, but endanger the foreign civilians on the ground. This practice reflects a concern of Western governments for their own soldiers. It also undermines the idea that everyone deserves equal protection (Archibugi 2004; Heinze 2006). The conflict between these values sits uneasy with us. We want clear answers where none can be found.

I think we can look at the AWS debate from a similar standpoint. AWS can promote human dignity in some ways – such as by making war more predictable and less cruel. They can undermine it – such as by expressing disrespect. In the end, it may not be an issue of which moral values are violated or not. At least not yet. Before we can launch the debate, we may ask which values we want to preserve, and which ones we can live without. Before we can talk about AWS, we should settle which acts truly express the moral concern that conscious beings deserve.

5. Conclusion

This essay has re-framed the debate on autonomous weapons systems as a question of non-conscious on conscious violence. I have argued that the dignity-argument against AWS relies on two facts: that humans are conscious and that machines are not (at least for now). I then argued that what we find so objectionable with autonomous weapons is non-conscious machines killing conscious humans. Unfortunately, I think that the debate cannot be resolved here. Before we can debate about AWS, we should clarify what we deem worthy of protection about conscious beings.

References

Amoroso, Daniele, and Guglielmo Tamburrini. 2018. “The Ethical and Legal Case Against Autonomy in Weapons Systems.” Global Jurist 18 (1). https://doi.org/10.1515/gj-2017-0012.

Amoroso, Daniele, and Guglielmo Tamburrini. 2020. “Autonomous Weapons Systems and Meaningful Human Control: Ethical and Legal Issues.” Current Robotics Reports 1 (4): 187–94. https://doi.org/10.1007/s43154-020-00024-3.

Archibugi, Daniele. 2004. “Cosmopolitan Guidelines for Humanitarian Intervention.” Alternatives: Global, Local, Political 29 (1): 1–21. https://doi.org/10.1177/030437540402900101.

Asaro, Peter. 2012. “On Banning Autonomous Weapon Systems: Human Rights, Automation, and the Dehumanization of Lethal Decision-Making.” International Review of the Red Cross 94 (886): 687–709. https://doi.org/10.1017/S1816383112000768.

Cutter, Brian. 2017. “The Metaphysical Implications of the Moral Significance of Consciousness.” Philosophical Perspectives 31: 103–30. https://doi.org/10.1111/phpe.12092.

Guizzo, Erico, and Evan Ackerman. 2016. “When Robots Decide to Kill.” IEEE Spectrum 53 (6): 38–43. https://doi.org/10.1109/MSPEC.2016.7473151.

Habermas, Jürgen. 2010. “The Concept of Human Dignity and the Realistic Utopia of Human Rights.” Metaphilosophy 41 (4): 464–80.

Hammond, Daniel N. 2015. “Autonomous Weapons and the Problem of State Accountability.” Chicago Journal of International Law 15 (2): 652–87.

Heinze, Eric A. 2006. “Humanitarian Intervention and the War in Iraq: Norms, Discourse, and State Practice.” The US Army War College Quarterly: Parameters 36 (1): 20–34. https://doi.org/10.55540/0031-1723.2291

Heyns, Christof. 2013. “Report of the Special Rapporteur on Extrajudicial, Summary or Arbitrary Executions.” UN Human Rights Council. https://primarysources.brillonline.com/browse/human-rights-documents-online/promotion-and-protection-of-all-human-rights-civil-political-economic-social-and-cultural-rights-including-the-right-to-development;hrdhrd99702016149.

Heyns, Christof. 2017. “Autonomous Weapons in Armed Conflict and the Right to a Dignified Life: An African Perspective.” South African Journal on Human Rights 33 (1): 46–71. https://doi.org/10.1080/02587203.2017.1303903.

Horowitz, Michael C. 2016. “The Ethics & Morality of Robotic Warfare: Assessing the Debate over Autonomous Weapons.” Daedalus 145 (4): 25–36. https://doi.org/10.1162/DAED_a_00409.

Kahane, Guy, and Julian Savulescu. 2009. “Brain Damage and the Moral Significance of Consciousness.” Journal of Medicine and Philosophy, no. 34: 6–26. https://doi.org/doi:10.1093/jmp/jhn038.

Königs, Peter. 2022. “Artificial Intelligence and Responsibility Gaps: What Is the Problem?” Ethics and Information Technology 24 (3): 36. https://doi.org/10.1007/s10676-022-09643-0.

Kriegel, Uriah. 2017. “Dignity and the Phenomenology of Recognition-Respect.” In Emotional Experience: Ethical and Social Significance, edited by J.J Drummond and S. Rinofner-Kreidl. Rowman & Littlefield.

Kriegel, Uriah. 2019. “The Value of Consciousness.” Analysis 79 (3): 503–20. https://doi.org/10.1093/analys/anz045.

Lee, Andrew Y. 2018. “Is Consciousness Intrinsically Valuable?” Philosophical Studies 176: 655–71. https://doi.org/10.1007/s11098-018-1032-8.

Lee, Andrew Y. 2024. “Consciousness Makes Things Matter.” Global Priorities Institute Working Paper Series, nos. 25–2024: 1–35.

Levy, Neil. 2014. “The Value of Consciousness.” Journal of Consciousness Studies 21 (1–2): 127–38.

Levy, Neil. 2024. “Consciousness Ain’t All That.” Neuroethics 17 (2): 21. https://doi.org/10.1007/s12152-024-09559-0.

Morrow, David R. 2014. “When Technologies Makes Good People Do Bad Things: Another Argument Against the Value-Neutrality of Technologies.” Science and Engineering Ethics 20 (2): 329–43. https://doi.org/10.1007/s11948-013-9464-1.

Mosakas, Kestutis. 2021. “On the Moral Status of Social Robots: Considering the Consciousness Criterion.” AI & SOCIETY 36 (2): 429–43. https://doi.org/10.1007/s00146-020-01002-1.

Nagel, Thomas. 1972. “War and Massacre.” Philosophy & Public Affairs 1 (2): 123–44.

Nagel, Thomas. 1974. “What Is It Like to Be a Bat?” The Philosophical Review 83 (4): 435–50.

Pape, Robert A. 2012. “When Duty Calls: A Pragmatic Standard of Humanitarian Intervention.” International Security 37 (1): 41–80.

Renic, Neil, and Elke Schwarz. 2023. “Crimes of Dispassion: Autonomous Weapons and the Moral Challenge of Systematic Killing.” Ethics & International Affairs 37 (3): 321–43. https://doi.org/10.1017/S0892679423000291.

Rosert, Elvira, and Frank Sauer. 2019. “Prohibiting Autonomous Weapons: Put Human Dignity First.” Global Policy 10 (3): 370–75. https://doi.org/10.1111/1758-5899.12691.

Sharkey, Amanda. 2019. “Autonomous Weapons Systems, Killer Robots and Human Dignity.” Ethics and Information Technology 21 (2): 75–87. https://doi.org/10.1007/s10676-018-9494-0.

Shepherd, Joshua. 2017. “The Moral Insignificance of Self‐consciousness.” European Journal of Philosophy 25 (2): 398–415. https://doi.org/10.1111/ejop.12221.

Shepherd, Joshua. 2018. Consciousness and Moral Status. 1st ed. Routledge. https://doi.org/10.4324/9781315396347.

Shepherd, Joshua. 2023. “Non-Human Moral Status: Problems with Phenomenal Consciousness.” AJOB Neuroscience 14 (2): 148–57. https://doi.org/10.1080/21507740.2022.2148770.

Shepherd, Joshua. 2024. “Sentience, Vulcans, and Zombies: The Value of Phenomenal Consciousness.” AI & SOCIETY 39 (6): 3005–15. https://doi.org/10.1007/s00146-023-01835-6.

Sparrow, Robert. 2016. “Robots and Respect: Assessing the Case Against Autonomous Weapon Systems.” Ethics & International Affairs 30 (1): 93–116. https://doi.org/10.1017/S0892679415000647.

UNIDIR. 2015. “The Weaponization of Increasingly Autonomous Technologies: Considering Ethics and Social Values.” https://unidir.org/wp-content/uploads/2023/05/considering-ethics-and-social-values-en-624.pdf.

Footnotes

1^ For a discussion of these points see Hammond (2015), Rosert and Sauer (2019), Amoroso and Tamburrini (2020) and Königs (2022).

2^ I thus ignore practical considerations, such as the fact that AWS may not be worse than human-controlled weapons One may argue that that most deaths in war by other people are just as undignified as death by robot (Horowitz, 2016). This is an interesting point if one asks if AWS should be banned or not. But this is not the question here.

3^ These weapons are already being developed (see Guizzo and Ackermann 2016).

4^ One may claim that the dignity of all humanity is violated by AWS (see Sharkey 2019 for a discussion). However, this position comes with some theoretical baggage. It would have to explain why my dignity is violated when someone on another continent is killed by a robot. In this essay, I focus on the dignity of the person killed by an autonomous system, not humanity as a whole. This also seems to be the point many critics of AWS address.

5^ There are many different aspects of consciousness and various approaches to its relationship with moral status. One candidate for the grounds of moral status is self-consciousness – the capacity to think of oneself as oneself and to understand ourselves as a continuous being in time and space. One argument states that only those who understand themselves as a continuing entity have a right to life. However, this approach has some issues (see Shepherd 2017).

6^ Another interesting debate is if consciousness itself has value (see Lee 2018 for a discussion).

7^ Recognition-respect is similar to Kant’s notion of respect, but Kriegel’s approach has a crucial difference. Where Kant grounds dignity in being an end-setting entity, Kriegel focuses on consciousness. He argues that non-conscious machines could, if sophisticated enough, be complex, end-setting beings, but they would not have moral status. By contrast, conscious beings that had no capacity to influence their surroundings – and would thus not be ends-setting – but possessed an interest in having certain phenomenal states, would have moral status (Kriegel 2017).

8^ It is unclear how far Kriegel’s account of dignity applies. Are we equally prohibited in automatically killing animals? It seems that the intuition is much weaker here – in any case, this is a task for another essay.

9^ A possible escape would be that all complex life forms have dignity because they are complex life forms. But this should still make us ask what unites them in having dignity.

10^ There may be some other considerations, such as how many agents must deliberate and which rights the person who is to be killed has. For instance, supporters of the death penalty may agree that it should not be applied without due process. For this essay, however, I am only concerned with the minimal requirement of having conscious agents decide over the death and not non-conscious ones.

11^ I talk of direct choices to kill or harm human beings. Algorithms already make decisions that might lead to the death of humans, but these are not direct. For instance, suppose that a desperate job seeker decides that, if he receives another rejection, he is going to commit suicide. Since algorithms already screen applicants, there may arise a situation where the algorithm rejects the applicant and he commits suicide. But in this case, the algorithm has not chosen if the job seeker is to live or not – it has ruled if he should be employed. Such situations are different from autonomous weapons.

12^ For a discussion on how all of humanity’s dignity may be violated see Sharkey (2019). Since discussions focus often on the dignity of the person killed, I will proceed with this assumption.

13^ See also Asaro (2012).