Abstract

As Artificial Intelligence transitions from static models to autonomous agents, we enter what Bari (2026) terms an "ASI-enabling distribution", a state where technical intelligence is ready, but systemic governance is absent. We argue that this transition, described as a "Pandemic of Agents," threatens a societal "bloodbath" characterized by rapid economic displacement and an accountability vacuum (Bari, 2026). In this paper, we address this crisis by integrating Sociotechnical Systems (STS) Theory, Stewardship Theory (Davis et al., 1997) , and the Formal Theory of Delegated Authority (Aghion & Tirole, 1997) to propose a new governance paradigm. We base our framework on two primary pillars. First, we introduce Recursive Alignment, utilizing the GPT(n-1) generation as a stable, automated auditor for GPT(n) model. Within an STS framework, we use this to create a technical "checks-and-balances" layer that allows the social system (humanity) to maintain oversight at machine speed. Second, we advocate for Operational De-Anthropomorphization. By applying the Theory of Delegated Authority, we argue that we must reclassify AI "Agents" as non-human delegates rather than digital personas. We believe this eliminates the "liability vacuum" where users blame the tool for its errors, firmly re-anchoring accountability to the human principal. Furthermore, we pivot from the constraints of Agency Theory to Stewardship Theory (Davis et al., 1997), which often views agents through a lens of self-interest and goal conflict (Aghion & Tirole, 1997). We posit that AI agents, devoid of intrinsic self-interest, should be engineered as "Algorithmic Stewards." In our model, the n-1 auditor serves as a repository of human-aligned values, ensuring that the more capable n model operates as a pro-organizational actor dedicated to collective utility rather than individualistic optimization. We contribute to the Information Systems (IS) community by reframing the "AI alignment" problem from a purely technical challenge to a fundamental IS Governance issue. We move beyond abstract ethics by providing a structural model for high-velocity accountability. By introducing "Recursive Stewardship," we offer IS researchers a theoretical bridge to study the co-evolution of human authority and machine autonomy, ensuring that the transition to super-intelligence is a controlled evolution rather than a societal shock.

Download

Share

COinS