TL;DR
The argument further holds that AI does not possess moral agency in the classical sense but functions as an infrastructural precondition for the reconfiguration of normative hierarchies—in an empirical rather than transcendental sense.
The rise of agentic AI systems, which are autonomous, proactive, and capable of multi-step task execution, has transformed how individuals interact with intelligent technologies. While these systems promise efficiency and enhanced decisionmaking, they also introduce new ethical vulnerabilities. This study investigates a paradoxical mechanism in AI-assisted academic task delegation: as social acceptance of AI delegation increases, individuals rely less on internal moral regulation. Drawing on Moral Disengagement Theory and Social Norms Theory, we test a normative substitution model using SEM data from 280 European university students and find that subjective norms function as both mediator and moderator, amplifying delegation intentions while reducing the influence of moral disengagement. Shame proneness emerges as a secondary moderator that buffers the normative pull for individuals with strong internal moral emotions. These findings highlight a critical socio-technical risk: proactive AI systems may unintentionally erode moral accountability as their use becomes socially normalized. We discuss implications for responsible agentic AI design, governance, and human-AI collaboration.
LLMs are increasingly used in morally sensitive contexts, yet it is unclear whether they apply ethical principles consistently across situations. A model that can state a moral principle may still violate it when the same scenario is rephrased or reframed. This inconsistency is a problem for any system whose outputs are used to inform moral decisions. If generative systems exhibit internal inconsistency, then the epistemic integrity of AI-mediated systems becomes uncertain. To study this concern, we investigate the stability of moral reasoning in LLMs within a controlled prompting framework across three major philosophical schools of thought: deontology, utilitarianism, and virtue ethics. We construct sets of morally equivalent scenarios in which the underlying situation is held constant while the framing varies to reflect different ethical stances and stylistic perturbations. We then evaluate responses from multiple models, including GPT, Mistral, and Llama. To assess consistency, we convert model outputs into structured logical statements and identify contradictions across responses generated within the same school of thought. Our results reveal substantial inconsistency with contradiction rates reaching up to 78% across scenarios. These findings point to a broader phenomenon of epistemic instability in generative AI wherein models fail to reliably maintain coherence with respect to their own prior outputs. This kind of instability carries real consequences. As generative systems influence how people form beliefs, judge actions, and absorb values, their inconsistencies can shape human reasoning and decision-making as well. Moreover, if a system cannot consistently represent its own normative commitments, then value alignment becomes a moving target rather than a well-defined objective. Thus, we argue that demonstrating internal incoherence is a necessary precursor to AI alignment.
Contemporary AI systems are increasingly designed to reduce uncertainty, ambiguity, and cognitive burden. This article argues that such systems do not merely expand human capacities but progressively externalize forms of judgment and existential burden that Christian traditions have historically understood as necessary conditions for the cultivation of moral agency. Drawing on the theological anthropologies of Augustine, Søren Kierkegaard, and Dietrich Bonhoeffer, the article develops a theological-anthropological critique of what it terms optimization culture: the normative logic through which the reduction of friction and existential difficulty becomes a dominant telos of technological design. Contemporary articulations of this logic are examined through the work of Jacques Ellul, Byung-Chul Han, Peter Thiel, Balaji Srinivasan, and William Davies. The article argues that AI systems shaped by this logic risk progressively externalizing the practices of disciplined attention, existential responsibility, and costly judgment through which mature moral agency has historically been cultivated. In response, it proposes a formative anthropology of technology as a framework for evaluating AI not only according to functional criteria such as safety or efficiency, but according to the kinds of human beings such systems actively help to form.
Contemporary institutions increasingly treat optimised procedures as decisions in their own right. This article advances an ontological limit claim for AI governance: moral judgment is constitutively personal and therefore non-delegable. Building on a minimal philosophical anthropology—person/thing distinction; irreducibility of phronēsis; the person as an end; responsibility as constitutive; and the capacity to initiate—we argue that algorithmic assistance can legitimately expand human deliberation, whereas delegation dissolves the very subject who judges. We situate the claim within current debates on artificial agency and Meaningful Human Control (MHC), and show how a subject-preserving reading supplements tracking/tracing by specifying what must remain human in dignity-touching domains. Two diagnostic cases—criminal-justice risk scoring and AI-steered coverage decisions in healthcare—illustrate a structural tendency to displace judgment by optimisation; they are not offered as empirical proof but as paradigmatic contexts where institutional deference to model outputs risks rendering answerability merely nominal. From our axioms we derive governance theorems for non-delegability, contestability, reversibility, and subsidiarity, and we close with an implication for the institutional formation of those who exercise judgment, which exceeds the scope of this article and is taken up in a companion paper. The result is a framework that welcomes instrumental progress while marking a principled boundary: systems may optimise for us; they may not judge instead of us.
The paper’s central argumentative shift is to change the narrative from bias mitigation to bias management—treating bias not as a defect to be corrected but as an ongoing condition to be governed.
This study conducts a technical analysis of frontier generative AI algorithms—including Meta’s Self-Rewarding Language Models, DeepMind’s EVA (Evolving Alignment via Asymmetric Self-Play) framework, and DeepSeek’s pure reinforcement-learning models—in order to examine an intrinsic paradigm shift in the ethical governance of generative artificial intelligence and to advance a physicalist analysis of algorithmic endogenous ethics. Combining a close reading of alignment techniques (RLHF, DPO, iterative DPO, GRPO) with a conceptual analysis grounded in Peter-Paul Verbeek’s theory of technological mediation and moral materialization, the paper traces how value-alignment goals are being “materialized” into internal, dynamic, and evolvable “moral scripts” within the algorithms themselves. The analysis shows that contemporary alignment practices are moving from external ethical discipline toward endogenous norms generated through iterative self-evaluation, asymmetric self-play, and rule-based self-exploration. The paper argues that this trend warrants a re-examination of Verbeek’s framework for its capacity to explain the co-evolution of technology and morality in the digital age, and it envisions a future of human–machine value co-evolution organized around new research directions such as “Setting as Governance” and “value homeostasis mechanisms”.