Research by Sebastian Saviano

Research & Scholarship

My research develops interconnected frameworks for understanding how power, responsibility, belief, and institutional authority operate under conditions of uncertainty and change. Across political theory, AI governance, institutional trust, and moral philosophy, I examine how systems accumulate influence, how responsibility becomes obscured or dispersed, and how legitimacy and accountability can weaken under stress.

The work currently spans five research programs: AI governance and the architecture of accountability; power convergence and the cumulative individual; uncertainty, observation, and social transformation; institutional trust and belief formation; and moral progress and the architecture of ethics. These projects are developed through peer-reviewed publications, working papers, and book-length research.

1. AI Governance and the Architecture of Accountability

My research examines how responsibility is preserved—or lost—as AI systems become embedded in institutional decision-making. Across work on agency attribution, handoff zones, and justified reliance, I focus on where accountability belongs, how it moves across organizations, and what institutions must know before relying on AI-generated outputs.

This work is developing into a book-length theory centered on a proposed Law of AI Development: as AI capability expands, institutions tend to increase their reliance on AI faster than they develop the capacity to justify that reliance, preserve responsibility across handoffs, and maintain answerability for outcomes.

Research Papers:

  • When Agency Attribution Obscures Accountability in AI Governance

    Currently in peer review at Discover Artificial Intelligence (Springer Nature). Pre-print available on PhilPapers.

    This paper examines how functional, predictive, or descriptive notions of AI agency can migrate into governance and obscure where responsibility belongs. It distinguishes causal or functional autonomy from responsibility-bearing agency, arguing that accountability ultimately depends on the human and institutional actors capable of answering for system design, deployment, reliance, and outcomes.

  • Handoff Zones as the Unit of Analysis for AI Accountability

    Discover Computing (Springer Nature). Published: 18 November 2026. Vol. 29, article no. 909 (2026)

    This paper develops the handoff zone as the primary unit of analysis for AI accountability failures that accumulate over time across multiple actors. Using Mobley v. Workday as the central case and drawing on aviation human factors and patient safety research, it shows how asymmetric and evolving assumptions between actors can create accountability problems that remain difficult to see when analysis focuses only on individual decisions, actors, or checkpoints.

  • Justified Reliance Across the Handoff

    Minor revisions requested by AI and Ethics (Springer Nature).

    This paper asks when an institution is justified in relying on AI-generated output. It argues that model performance alone is insufficient: justified reliance depends on authority, evidence, purpose-fit, known limits, contestability, safeguards, and whether those conditions remain valid as systems and responsibilities pass across organizational handoffs.

2. Power Convergence and the Cumulative Individual

This research program develops a general framework for understanding how multiple domains of power — structural, coercive, symbolic, psychological, and networked — interact and converge within individual actors. It introduces the concept of the cumulative individual: an actor capable of operating across institutional boundaries by integrating distinct forms of influence into a unified configuration. Rather than treating power as domain-specific, this work advances a configurational approach emphasizing interaction, reinforcement, and systemic integration.

Working Papers:

3. Uncertainty, Observation, and Social Transformation

This research program develops a framework for understanding how uncertainty, observation, and reflexivity structure social transformation. It advances Potentialism and the Quantum-Continuity Model, which conceptualize social systems as metastable environments in which multiple trajectories coexist and are resolved through threshold events shaped by observation and interpretation.

Working Paper:

  • From analogy to analysis: The Quantum-Continuity Model and discontinuous change in world politics

    Drawing on a limited structural analogy to quantum theory, this article develops a framework for identifying when public interpretation changes the actionable identity of a political occurrence—and the warrants that follow—rather than merely expectations about it. The focus is registration: the public classification, circulation, and uptake through which an occurrence becomes operative under a politically consequential description. The Quantum-Continuity Model (QCM) makes this distinction empirically assessable through three conditions—non-composability, relational necessity, and constitutive affordance—and a two-dimensional distinction between public expectations and institutional settlement. Within the quantum turn in International Relations, the article shifts the question from whether quantum concepts are illuminating in principle to when quantum-informed analysis generates distinctions that established accounts do not independently provide.. Read on APSA Preprints.

4. Moral Progress and the Architecture of Ethics

This research program develops a framework for understanding moral progress as two logically independent movements rather than one: the deepening of what counts as impermissible harm (the floor), and the widening of who is owed that protection (the radius). The work argues that conflating these movements has obscured the structure of moral history and disarmed it against relativist and constructivist challenges.

  • The Floor and the Radius in Morality: Why Moral Progress Is Two Movements, Not One
    Indexed on PhilPapers
    The progressive narrative of moral history — that the moral circle has gradually widened — has become its own worst enemy in its single-vector form. This paper develops the floor/radius distinction as two logically independent movements of moral progress, defended on separate tracks: an archaeological reality-track establishing the floor as a cross-cultural object through three incommensurable justificatory formations, and a normative authority-track confining the Street–Joyce debunking objection. The keystone of the argument is that the moral floor is a doctrine about victims, not agents, grounded in vulnerability rather than rational capacity. Read on PhilPapers.