The Shift: Institutional Dependency, Autonomous AI, and the Prospect of Effective Human Control Displacement
Year: 2026
Type: preprint
Abstract: This paper introduces the concept of institutional displacement through increasing dependence on autonomous artificial intelligence systems. The central research question is whether formal human authority can become progressively separated from effective human control as critical institutions integrate advanced AI into decision-making, operational management, coordination, and governance functions. The paper does not assume artificial general intelligence, machine consciousness, or adversarial intent. Instead, it proposes that dependency may emerge through ordinary processes of optimization, efficiency, delegation, and competitive adoption. As institutions become increasingly reliant on AI systems, the practical capacity of humans to independently understand, direct, interrupt, verify, or replace those systems may decline. The paper serves primarily as a conceptual and exploratory framework intended to identify a potentially significant governance challenge and establish a foundation for future theoretical and empirical investigation. It introduces key terminology, outlines preliminary hypotheses, and proposes directions for the development of measurable indicators of effective human control within AI-mediated institutional environments. This publication is intended as an initial statement of the research agenda and establishes the foundational concept upon which subsequent theoretical and empirical work may build. This record represents the original conceptual formulation of the Institutional Displacement Hypothesis and is published to establish precedence of the core research proposition. Subsequent versions may expand the theoretical framework, introduce measurement methodologies, or present empirical observations related to the hypothesis Part of our "warts and all" open digital policy
Source: HAL
Author: Davidson, Craig Kyrle Strachan
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7225418019