Danielle Marie Swanepoel

Determining the agency-status of machines and AI has never been more pressing. As we progress into a future where humans and machines more closely co-exist, understanding hallmark features of agency affords us the ability to develop policy and narratives which cater to both humans and machines. This paper maintains that decision-making processes largely underpin agential action, and that in most instances, these processes yield good results in terms of making good choices. However, in some instances, when faced with two (or more) choices, an agent may find themselves with equal reasons to choose either - thus being presented with a tie. This paper argues that in the event of a tie, the ability to create a voluntarist reason is a hallmark feature of agency, and second, that AI, through current tie-breaking mechanisms does not have this ability, and thus fails at this particular feature of agency.

Technological unemployment is a very real phenomenon that should be addressed by governments and businesses alike. This paper argues that current approaches to technological unemployment are short-sighted in that they focus predominantly and primarily on current generations. This kind of approach results in harm such as ignoring impending meaning-crises and propagating a potential form of human-quota-driven tokenism in the process of implementing automation in the workplace. Arguably, current generations can (and should) benefit from communal resources insofar as they do not harm the least privileged of future generations. That is the threshold that is set in this paper and this threshold can serve as a guiding principle for leaders when making decisions about whether automation should be implemented, the degree and scope to which automation should be implemented, and the policies regarding the process of the implementation and the regulation of automation in the workplace. An intergenerational justice approach is a novel approach to the problem of technological unemployment and can inform decision-making and policy-setting alike.

To be said to possess agency, an individual or entity must act with intent. What constitutes an intentional action is, roughly, that the action is something an agent wishes or desires to do. In order to perform intentional actions however, there seems to be much that must underpin this process. For example, having a complex range of desires, beliefs, and deliberations which are involved in bringing those intentional actions about. A satisfactory account of agency should be able to explain how these complex range of desires, beliefs and deliberations occur to bring about intentional actions. The aim of this paper is to answer whether Artificial Intelligence has what it takes to be considered agents. At first glance, it appears that AI can calculate, weigh up different options, choose the best one which will ensure the best possible outcome, right a wrong, troubleshoot, solve a problem etc. And, we might further be moved to say that these machines perform actions, perhaps we could go so far as to say that they seemingly perform intentional actions. But are they agents? As things stand, according to the features of common-ground agency, AI in its more traditional rules-and-representations version, and possibly, AI in its more dynamic version, fail to exhibit the features of common-ground agency and therefore fail at the game of agency.

Moral agency status is often given to those individuals or entities which act intentionally within a society or environment. In the past, moral agency has primarily been focused on human beings and some higher-order animals. However, with the fast-paced advancements made in artificial intelligence, we are now quickly approaching the point where we need to ask an important question: should we grant moral agency status to AI? To answer this question, we need to determine the moral agency status of these entities in society. In this paper I argue that to grant moral agency status to an entity, deliberate norm-adherence must be possible. In this paper I argue that, under the current status quo, AI systems are unable to meet this criterion. The novel contribution this paper makes to the field of machine ethics is first, to provide at least two criteria with which we can determine moral agency status. We do this by determining the possibility of deliberate norm-adherence through examining the possibility of deliberate norm-violation. Second, to show that establishing moral agency in AI suffer the same pitfalls as establishing moral agency in constitutive accounts of agency.

The knowledge argument is an argument for dualism that claims that there are both physical and non-physical facts, something we can know by reflecting on 'Mary' who is aware of all scientific data about colours but has yet to see any. I reject the dualist conclusion and instead provide a new physicalist response that I call 'bifactualism'. Bifactualism is a novel physicalist account comprising two elements. First, like dualism, it distinguishes between two kinds of facts: general and particular facts. Second, unlike dualism, it claims that the general/particular distinction may explain facts about experience. There are certain facts that go undocumented in what is expressible in the language of the physical sciences because the language of the physical sciences concerns only general facts, whereas experience, I argue, essentially involves learning particular facts. Thus, I argue that the case of Mary does not support mind/body dualism, and instead provides at least equal reason to support bifactualism. Since the general/particular distinction is one we are stuck with regardless of the status of mind, bifactualism emerges as a more parsimonious and hence preferable account of experience.