Jovana Davidovic

The increased use of AI for warfighting has given rise to worries about the so-called responsibility gap, namely the worry that when things go wrong (e.g., a disproportionate number of civilians are killed by an AI-enabled weapon) we might not know who to hold responsible for that failure or, more precisely, that no individual or group involved in the AI system’s life cycle (from design to deployment) may meet the conditions for attributing responsibility when failures occur. This has led some scholars to argue that such weapons (ones that have a responsibility gap) cannot be justifiably deployed. This paper, following recent work in the field, challenges the assumption that the responsibility gap is a singular, univocal problem, arguing instead that multiple, distinct responsibility gaps exist, and that they can be categorized along different conceptual, meta-theoretical, and normative lines. The paper specifically argues that the meta-theoretical/pre-theoretical question is essential and that the first question we need to ask is what is the purpose of trying to attribute responsibility for a particular outcome or action—is it so that we can assign blame, justify punishment, determine liability to defensive harm, ground accountability of some type, or something else. Being clear about the purpose of assigning responsibility can profoundly shape how we understand, and ought to respond to, potential responsibility gaps. Ultimately, the paper argues that the justified deployment of warfighting AI requires understanding which types of gaps are relevant, why they matter, and how they might be addressed through tailored socio-technical implementation mechanisms.

Horizon scanning is intended to identify opportunities and threats associated with technology, regulatory, and social change. Here, we report the results of a new horizon scan based on inputs of an international group of 33 participants, focusing on future issues arising from the military use of artificial intelligence (AI) for augmenting human performance. The final list of 12 issues includes topics spanning from the political (educating and training individuals to accept and work with AI), to the regulatory (issues of consent to human-AI teaming and hybridization), to security (the hackability of neural devices that connect to AI), to philosophical (the nature and phenomenology of brain-to-brain interfaces). The early identification of such issues is relevant to researchers, policymakers, military practitioners, and the wider public.

Horizon scanning is intended to identify opportunities and threats associated with technology, regulatory, and social change. Here, we report the results of a new horizon scan based on inputs of an international group of 33 participants, focusing on future issues arising from the military use of artificial intelligence (AI) for augmenting human performance. The final list of 12 issues includes topics spanning from the political (educating and training individuals to accept and work with AI), to the regulatory (issues of consent to human-AI teaming and hybridization), to security (the hackability of neural devices that connect to AI), to philosophical (the nature and phenomenology of brain-to-brain interfaces). The early identification of such issues is relevant to researchers, policymakers, military practitioners, and the wider public.

Bio-convergent enhancements for soldiers are becoming increasingly inevitable. Medical professionals, bioethicists, lawyers, and neuroscientists are increasingly aware of the potential for these enhancements to raise significant ethical issues, especially around issues of consent and responsibility for long-term care. This has, in the last few years, led to an increase in research on the ethics of soldier enhancements. The literature on this issue has rightly leveraged decades of bioethics, medical ethics, and research ethics literature. What is missing however from the literature is the perspective of the potential subjects of such enhancements, namely members of special operations forces. This paper seeks to fill this gap, by first arguing that subjective views of special operations members matter for ethical questions and then by reporting results of our interview-based qualitative study on United States Special Operations Forces’ perspectives on consent and long-term care.

In this paper I focus on duties we owe refugees from conflict zones. I argue that it is important to distinguish between two types of duties one might have with respect to refugees from conflict zones. Belligerents from wars that resulted in excess numbers of refugees, I argue, have a stringent duty to remedy past harms and provide for resulting refugees. Other states have a duty to aid which is context-dependent and can be in some cases as stringent as the duty to remedy past harms. I argue that making a distinction between the grounds and types of duties different actors have with respect to refugees from conflict zones has significant consequences both for just war theory, but also for practical questions about how best to discharge our duties to refugees from conflict zones.

I argue here that certain species of war, namely humanitarian military interventions (HMIs), can be obligatory within particular contexts. Specifically, I look at the notion of HMIs through the lens of just war theory and argue that when a minimal account of jus ad bellum implies that an intervention is permissible, it also implies that it is obligatory. I begin by clarifying the jus ad bellum conditions (such as just cause, right intentions, etc.) under which an intervention is permissible. I then turn to the claim that permissibility necessitates obligation, by first showing that whenever an intervention is permissible, it is also minimally decent. Second, I show that minimally decent actions are morally obligatory by arguing that the notion of minimal decency is a conceptual bridge between negative and positive duties. Third, I argue that performing minimally decent actions is necessary for a state to be just. Ultimately, my conclusion arises from the following observation: if a humanitarian crisis is bad enough for one to hold that it is permissible to breach sovereignty of a nation, then it is bad enough to hold that there is an obligation to intervene.

Meaningful human control over AI is exalted as a key tool for assuring safety, dignity, and responsibility for AI and automated decision-systems. It is a central topic especially in fields that deal with the use of AI for decisions that could cause significant harm, like AI-enabled weapons systems. This paper argues that discussions regarding meaningful human control commonly fail to identify the purpose behind the call for meaningful human control and that stating that purpose is a necessary step in deciding how best to institutionalize meaningful human control over AI. The paper identifies 5 common purposes for human control and sketches how different purpose translate into different institutional design.