An attentional injustice occurs when some person or group is given disproportionate attention. In academic philosophy, it has been observed that members of various marginalised groups are cited less frequently than white men, both in academic papers and in university syllabi. This has downstream effects on gender/racial representation in Philosophy more generally. Because Large Language Models produce text due to prediction on the basis of training data, they are likely to replicate existing biases. As a result, without precautionary measures, LLM use is likely to exacerbate attentional injustice in philosophy and contribute to the continuing of current demographic distortions.
Editorial policies increasingly require authors to disclose substantive uses of generative AI, but such requirements commonly focus on whether, where, and how a model was used. Even function-specific disclosure may locate a model in the research or writing process without showing how its outputs entered scholarly judgment, how they were evaluated, or whether contributors can reconstruct the relevant reasoning and defend resulting claims without treating those outputs as sufficient justification. Drawing on research ethics, AI disclosure policy, human–model interaction, and researcher formation, this article argues that disclosure documents model use, whereas responsible use requires claim-level authorial warrant: a contributor’s capacity to understand, justify, and defend claims within their contribution. The article distinguishes cognitive delegation, in which model outputs shape substantive choices, interpretations, or justifications, from routine delegation, which does not ordinarily bear on the substance of judgment. Its conceptual contribution is to link the epistemic role of outputs that shape judgment to the function-appropriate evaluation required to justify decisions and retain warrant for resulting claims. It then repurposes disclosure categories as a claim-sensitive ex ante heuristic, mapping six recurrent research functions to risks, forms of oversight, and formative questions for recognising cognitive delegation and selecting function-appropriate evaluation. The article contributes to the ethics of information technology by clarifying when AI-supported scholarship remains under responsible human judgment and when cognitive delegation threatens the epistemic basis of authorship.
In philosophical work on the ethics of video games, ludic amoralists argue that acts players perform in games are beyond moral scrutiny. This is chiefly supported by the following arguments: the fictional harm argument, the consequentialist rebuttal, and the moral exploration argument. The fictional harm argument supports ludic amoralism in principle: if all game actions are fictional, none can be immoral. The consequentialist rebuttal supports a defeasible default: if violent games do not reliably produce harmful outcomes, amoralism is the default position. The moral exploration argument provides an aesthetic reason for amoralism: play is beyond moral criticism insofar as the player engages with the game in an aesthetic, exploratory way. Drawing on Waltonian theories of fiction, I argue that players often imagine themselves as performing the fictional acts represented in games, ranging from heroic exploits to morally abhorrent deeds. When players do this, they can come to identify with, to varying degrees, their fictional exploits. Strong identification occurs when players interpret their fictional actions as satisfying real desires, such as the desire to harm others. Using examples from players of violent extremist games, as well as mainstream violent games, I show how such identification can arise in practice. Most importantly, I show that in cases of strong identification, all three justifications for ludic amoralism fail.
Artificial intelligence systems increasingly do more than influence human decision-making; they constrain, block, suspend, and sometimes bypass it. Fraud-detection systems halt transactions, autonomous vehicles override steering and braking, platforms suspend accounts, and emerging neurotechnologies may modulate affective or cognitive states upstream of action. Existing AI ethics has focused primarily on influence, manipulation, fairness, and accountability, but these frameworks do not adequately capture cases in which AI interferes directly with the pathway from a user’s reasons to their action. In this paper, I introduce the concept of digital restraints and argue that their ethics can be illuminated by analogy to the more developed ethics of medical restraints. Drawing on this model, I develop a taxonomy of the moral situations to which restraints respond, and an agency-protective account of permissible digital restraints organized around four conditions: decision-appropriateness, authorization, least agency-restricting means, and heightened scrutiny for reason-bypass. I also extend the dominant model of decisional capacity in medicine to develop an account of decisional capacity in digital contexts, arguing that digital environments can themselves degrade understanding, appreciation, reasoning, and expression. Because authorization is central to the permissibility of restraint, I introduce the concept of Advance Algorithm Directives (AADs) as a mechanism by which users may authorize future AI interventions under specified conditions. The result is a scalar framework for evaluating AI-mediated override across domains that permits AI override while protecting human agency.
Academic publishing has largely settled one question about generative AI while leaving another unresolved: AI systems are not eligible for authorship, but the boundary between legitimate assistance and authorship-displacing delegation remains unclear. This article argues that these are separate normative questions. Excluding an LLM from the byline does not by itself justify restricting human use to grammar, readability, or other non-intellectual tasks. At the same time, human approval and responsibility are not sufficient for authorship when authorship-defining intellectual work has been delegated. The relevant boundary is therefore between legitimate substantive assistance and authorship-displacing delegation. In conceptual scholarship, legitimate AI-assisted authorship requires a substantial authorship-relevant human intellectual contribution, critical and integrative command of the developing argument, verification and provenance discipline, proportionate disclosure, and final accountability. The article uses the speechwriter dilemma, peer review, philosophical dialogue, genealogy, and extended cognition to clarify why substantial non-author contribution need not displace human authorship, while still preserving the credit function of the byline. It concludes with a three-level governance model and applies the account reflexively to the article’s own AI-assisted revision process.
