Filed under:
Human Centricity
Understanding

Respecting the autonomy of those affected by any AI-enabled system is part of the Human Centricity principle. However, here we consider its relationship with Understanding with respect to those military personnel who may be directly connected with the system. Many of the same concerns will be directly relevant to other stakeholders as well if you change the specific example. To what extent should service members be able to decide whether to engage with human-machine teaming systems that the military has developed? This is a question about the role of informed consent.

1. Informed Consent & Autonomy
While consent and autonomy may be considered limited in some military contexts, personnel who are involved in human-machine teaming for example, should be fully informed about the implications of their interactions as far as is possible. An individual granting consent without being given sufficient information cannot really be said to have understood what they are consenting to, meaning the consent is unlikely to be considered valid. This is particularly the case where the AI-enabled system is part of a wider element of Human Augmentation. 

Autonomy is necessarily circumscribed for service members for much of their service life. Enlisting means accepting less control over one’s own life. On its own, this difference does not mean that autonomy and informed consent are irrelevant in the military. There are ethical limits to what service members can be expected to participate in. For example, there is (now at least) a consensus that conducting involuntary medical experiments on service members is unethical. While determining the precise amount of permissible autonomy limitation for service members is a complex issue, some interventions are clearly off limits, and a principle of least restrictive means ought to apply. Other values also apply. In all contexts, respect and dignity require, as much as possible, that people be free to make decisions for themselves. Decisions about what happens to our bodies are especially important. 

 
2. Informed Consent In Bioethics
In Bioethics the conditions for valid informed consent are as follows: 

 

Capacity: While capacity is a significant factor in medicine, capacity issues will be far less common for service members who are less diverse as a group than the population as a whole. That leaves information, understanding, and voluntariness, which we will consider in turn. 

 

Information: In bioethics and the law, most agree that consent is only valid if the individual is given all relevant information. 

These issues have military analogues. Providing the same level of information to service members about augmentation might be more difficult, since the details of the technology might be restricted. Nevertheless, providing information should be the default, and should only be limited when there is strong reason to do so. Service members can also be given general information when they enlist about the ways that their autonomy might be limited. Just as they should understand that they can be deployed around the world, if it is possible to inform them in general terms that augmentation is a possibility, this should happen. As with medicine, it is an open question whether service members should be able to waive their right to information. 

Understanding: It is not enough to be given information; the decision-maker must understand it. The justification for this condition is that understanding is necessary for the individual to decide if the intervention is acceptable to them all things considered (i.e., if it aligns with their values). In many places, it is all too common for patients to be given multi-page consent documents filled with medical and legal jargon. Instead, patients must be given the information in a way they can understand and demonstrate that they have understood it. A common method to measure understanding is to have the patient repeat the content back to the healthcare provider, who can then gauge the level of understanding. Importantly, since consent is decision-specific, each decision comes with its own required level of understanding. In medicine, a child or someone experiencing cognitive decline might be able to consent to simple interventions, such as a vaccination or what to eat that day, but be unable to understand complex surgery. 

There is an ethical case for setting a similar standard for service members consenting to augmentation. Plausibly, service members should understand what is being done to them and how it will affect them during their service and, if applicable, for the rest of their lives. They should also understand that unknowns exist for new technologies. 

 

Voluntariness: The voluntariness condition is both important and difficult to specify. Valid consent seems to require a certain level of voluntariness. When a mobster makes someone “an offer he can’t refuse,” implying that harm will come to him if he refuses, intuitively, this does not count as valid consent, since the person was coerced. But other cases of coercion, pressure, and manipulation less clearly invalidate consent. If a patient’s spouse says, “get this surgery or I’m leaving you,” the spouse is changing the decision-making calculus, but (it seems) the patient can still decide if he values the marriage or the lack of surgery more. That is, he is still free to decide as he wishes. The same goes for a manager saying “do this project or you’re fired” or even “get this vaccine or you’re fired.” Most people, including bioethicists, believe that there is a point where consent is invalidated because of involuntariness, but demarcating voluntary from involuntary decisions is difficult. 

Power imbalances exist in both medicine and the military. In both contexts, people can feel undue pressure to make a decision in order to please someone else, such as their doctor, commander, or fellow service members. Similarly, patients and service members might feel that they lack an actual choice. Steps should be taken to prevent this from happening in the military. Where possible, service members should be given a free choice and it should be clear to them that they have a free choice. 
 

3. Promoting Informed Consent
Autonomy, dignity, and respect all require, where possible, that service members be given complete information and be allowed to consent to or decline any intervention. This should be the default. However, there might be ethically acceptable circumstances where complete information or consent is infeasible. Identifying these situations requires thorough ethical consideration. Even when they are found to be ethically justified, there should be a standard of giving as much information as possible and involving the service member in the decision- making process. This is also analogous to medicine. Even if a patient is incapable of giving consent, their preferences should still be sought and they should be informed of their medical condition and proposed interventions where possible. Informed consent is not a binary, on-or-off process. Instead, the various elements can be considered and promoted independently. (See Case Study: Remote Monitoring of Military Divers – Determining Appropriate Control & Understanding Informed Consent)

While the example above outlines the issues relating to informed consent concerning human-machine teaming or even human augmentation, the broader ideas about what informed consent really means are pertinent to a much wider range of issues.