A Game-Theoretic Assessment of Optimal Strategies in Artificial Intelligence Competition Between the United States and China

Abstract

Semiquantitative analysis was done to ascertain the extent of the difference between the United States and China. This paper analyzes the competition between the United States(US) and China in artificial intelligence(AI) and uses game theory to examine its impact on US-China economic relations. As China-US AI competition heats up, this paper analyzes six factors of current AI competition between China and the US – AI models, semiconductors, researcher talent, AI patents, AI investment, and infrastructure – using time-series graphs to diagnose differences in overall capability. We hypothesized that, owing to the US leading, the US’s best strategy was to restrict China, and China’s best strategy was to not restrict the US. Using the framework of a normal-form game and looking at past and current strategic moves from both sides, this paper created a model to predict future moves from both sides. The paper also incorporates a Stackelberg model to evaluate strategic interactions in AI competition. By combining current AI competition data with game-theoretic analysis, this paper evaluates the likely effects of AI competition on future US-China economic relations and attempts to predict future behavior from both actors’ governments using mathematical modeling. This analysis reveals that Chinese and US decisions to de-escalate from mutual restriction were rational in the context of long-term relations, and that neither side has a stable strategy, with the hypothesis only being supported when both sides are cooperating.