Artificial General Intelligence and the Social Contract: A Dynamic Political Economy Model

This paper develops a dynamic model of Artificial General Intelligence (AGI) capital accumulation and explores its implications for long-run economic stability, human labor, and the viability of the social contract. Extending baseline growth models, we introduce fixed and variable costs of AGI scaling, classify these costs, and analyze their impact on steady-state outcomes. We prove that sublinear costs allow unbounded AGI accumulation, ultimately driving wages and employment to collapse, while superlinear costs impose endogenous limits that preserve human economic relevance. Building on this foundation, we model redistribution and bargaining between human agents and AGI capital owners as a dynamic game, demonstrating the existence of stationary redistribution equilibria that stabilize welfare in the presence of AGI. However, the analysis reveals that excessive political concentration or unforeseen technological shocks can destabilize these contracts, endogenously leading to welfare bifurcation or collapse. We extend classical social contract theory to this novel context, arguing that in AGI-dominated economies, sustainable social contracts must be dynamically incentive-compatible for both human and artificial agents. The results show that without adaptive institutional mechanisms and explicit redistribution, AGI expansion threatens to sever economic reciprocity, erode human welfare, and destabilize macroeconomic and political equilibrium. Thus, the emergence of AGI necessitates not only technological governance but a reconceptualization of the social contract itself.