Theory Workshop -Fulin Guo (Cambridge)

Title: “GPT in Game Theory Experiments”

Abstact: This paper explores the use of Generative Pre-trained Transformers (GPT) in strategic game experiments, specifically in the finitely repeated play of the ultimatum game and of the prisoner’s dilemma. I designed prompts and architectures to enable GPT to understand the game rules and to generate both its choices and the reasoning behind decisions. The key findings show that GPT exhibits behaviours similar to those of humans in important aspects, such as making positive offers and rejecting unfair ones in the ultimatum game, along with conditional cooperation strategy in the prisoner’s dilemma. The study explores how prompting GPT with traits of fairness concern or selfishness influences its decisions. Notably, the “fair” GPT in the ultimatum game tends to make higher offers and reject offers more frequently compared to the “selfish” GPT. In the prisoner’s dilemma, the “fair” GPT has a significantly higher cooperation rate than the “selfish” GPT. The reasoning statements produced by GPT during gameplay uncover the underlying logic of certain intriguing patterns observed in the games. They also reveal that GPT’s cooperative behaviours in the prisoner’s dilemma are sometimes driven, at least in part, by errors in reasoning. Overall, this research shows the potential of GPT as a valuable tool in social science research, especially in experimental studies and social simulations.