The possible emergence of artificial general intelligence (AGI) promises both awesome opportunities and not-yet-quantified risks. The arrival of new paths to AGI capabilities has been considered analogous to the development of nuclear weapons. Just as game theory emerged as a means to model and derive strategic equilibria during the early development and operations of nuclear weapons, the author proposes to use game theory to capture and formalize some of the strategic dynamics surrounding the paths to AGI.
In this paper, the author models the geopolitical race toward AGI as a series of strategic interactions between two primary actors—the United States and China. Beginning with a single-shot Prisoner’s Dilemma formulation, the author shows that, although mutual cooperation yields the highest economic growth, the dominant equilibrium is a mutual sabotage policy. Subsequent models incorporate certain factors: (1) the advent of destabilizing “wonder weapons” associated with AGI breakthroughs, (2) repeated-game dynamics, and (3) the potential for preemptive strikes, with each factor exacerbating competitive incentives and strategic instability. The author ultimately extends the model to include cumulative existential and systemic risks—such as misuse, misalignment, or proliferation—which could create conditions for cooperation supporting nonproliferation.
The author gives a play-by-play account of each game formulation, providing mathematical statements and player insight to determine various strategies’ most likely actions and payoffs. The author’s findings can help decisionmakers and general audiences, particularly those interested in examining the geopolitical race toward AGI, better understand the involved risks and potential pathways to cooperation
This publication is part of the RAND expert insights series. The expert insights series presents perspectives on timely policy issues.
This document and trademark(s) contained herein are protected by law. This representation of RAND intellectual property is provided for noncommercial use only. Unauthorized posting of this publication online is prohibited; linking directly to this product page is encouraged. Permission is required from RAND to reproduce, or reuse in another form, any of its research documents for commercial purposes. For information on reprint and reuse permissions, please visit www.rand.org/pubs/permissions.
RAND is a nonprofit institution that helps improve policy and decisionmaking through research and analysis. RAND’s publications do not necessarily reflect the opinions of its research clients and sponsors.
