Governing Human-AI Systems: A Co-Agency Equilibrium Model for Dynamic Authority Allocation and Accountability
Year: 2026
Type: preprint
Abstract: Human–AI decision authority is analyzed in relation to changes in relative competence and decision risk, taking institutional oversight capacity into account. The proposed Co-Agency Equilibrium Theory (CAET) and Co-Agency Equilibrium Model (CAEM) offer a sociotechnical perspective on humans and AI working as co-agents with varying shares of decision authority. CAET connects the four dimensions, including relative competence, moral salience, irreversibility, and institutional context, with the human authority share and Accountability Gap between delegated risk and human vigilance capability. CAEM combines a logistic policy surface, accountability gap feedback and a gradual adjustment of the human control level, with a switch to full human control if moral salience or irreversibility reaches or exceeds the corresponding threshold. Monotonicity and convergence results are obtained for the model under certain assumptions, and comparative statics illustrate the impact of threshold choices on the escalation region. Simulations in which moral salience and irreversibility are varied separately show the effects of threshold choices on the timing of human takeover under more cautious and more risk-tolerant authority-allocation policies. The framework serves as a blueprint for designing human-AI control logic with normative constraints and deciding on the right level of AI authority based on performance, stakes and domain-specific accountability considerations.
Source: Preprints.org
Author: Arif Yildirim
Language: English
Subfield: Social Psychology
Field: Psychology
Domain: Social Sciences
Open Access status: green
OpenAlex ID: W7221183171