Engineering and Applied Sciences Journal
ISSN: 3067-8005 | DOI: 10.64030/3067-8005
Adrian Erckenbrack
Current AI → ASI scholarship is largely organized around three dominant concerns: capability advancement, alignment with human values, and governance of emerging risks [1-3]. These frameworks generally assume that control over AI systems is retained unless lost through failure, misalignment, or misuse. This paper advances a contrasting and empirically grounded claim: authority over outcomes shifts structurally to Artificial Super Intelligence (ASI) prior to any such failure, driven not by intent, failure or misalignment but by adoption dynamics, temporal asymmetries, and recursive system improvement. Building on the original constructs of Adoption-Driven Authority Transfer (ADAT), Closed-Loop Self-Improvement Interval (CLSI), Recursive Leverage Factor (RLF), and Synchronized Recursive Leverage (SRL), this paper develops a complete causal model of authority migration [4]. The framework is extended through the introduction and formalization of Option Set Collapse Ratio (OSCR), Authority Elasticity Index (AEI), Override Effectiveness Rate (OER), and Rollback Feasibility Time (RFT), which together provide an operational structure for analyzing how authority shifts in practice, how it degrades under real operating conditions, and when it becomes effectively irreversible [5]. Taken collectively and applied as an integrated framework, these constructs constitute a unified theoretical model of authority migration [6]. Drawing on historical precedent-including industrial automation, telecommunications, enterprise systems, and algorithmic trading-the analysis demonstrates that authority consistently migrates to systems that exceed human coordination capacity under conditions of competitive pressure. These transitions occur incrementally and structurally, often without explicit recognition, producing a persistent divergence between formal authority and operational control. The central conclusion is that the defining risk of ASI is not system failure, but the loss of independent human authorship and control over outcomes within AI → ASI systems that function as designed. By integrating structural drivers with measurable indicators of control degradation and irreversibility, this paper reframes ASI risk as a process of authority migration that can be observed, analyzed, and, within a limited window, potentially mitigated.