Overview
My current research asks how we ought to conceive of AI-mediated causal chains, and applies a diagnostic method drawn from my doctoral work on William James to answer it. In a Jamesian account of action, an ascription of agency is legitimate only when an observer’s inference that behaviour felt like action is matched by the conditions for it having done so; this is why dreamed actions are not actions. My method is thus to take a marker we typically treat as evidence of genuine agency, identify the condition that makes that marker normatively significant, and ask whether AI systems which display the marker of agency are able to satisfy the condition for normative significance. In short, I check whether the thing we are reading as agency is backed by what is supposed to back it, and then consider what follows normatively from any answer.
What interests me is that the same approach keeps yielding different answers about whose problem this is, and whether any remedy could be adopted voluntarily or would have to be enforced through regulation. Working that out is the project’s next theoretical problem.
Most Recent Projects
How to Survive Online
This is a textbook on digital ethics aimed at lower-level undergraduates and community college students. It covers many topics of vital importance for being an informed and conscientious online citizen, such as: online relationships, influencer culture, algorithmic content delivery, trolling & ragebaiting, misinformation & propaganda, and more.
This book is under contract (Broadview Press), due at the end of 2027.
Paper 01 – AI and action theory
In this paper, I argue that all AI-mediated actions are necessarily deviant, either practically (weak case) or conceptually (strong case). Given the source of this deviance, the onus is on those who want to introduce AI systems into morally significant situations, and not those who wish to restrict it.
This paper is currently under consideration for publication.
Paper 02 – AI and effort
In this paper, I argue that inferring reliability from effort is warranted in the human case, but not in the AI case. The effort setting found in AI systems equivocates on the term ‘effort’ to give an illusion of agency, where none exists. This is especially problematic for those already in a poor epistemic position to judge an AI system’s output.
This paper is currently under consideration for publication.
Paper 03 – Petitionary prayer and AI use
In this paper, I argue that a user crediting a good outcome to an AI system’s judgement occupies the same epistemic position as the petitioner in religious epistemology in that both move from the fact that the good occurred to the claim that it was granted, with little to say about the steps in-between. The practices that tradition developed (e.g. habits of humility, institutional frameworks of discernment) serve as a model for how reliance on AI systems might be disciplined.
This paper will be submitted for review in early 2027.
