How artificial intelligence can reduce selfish behavior and reshape society

The study from Michigan State University explores the impact of artificial intelligence (AI) on human cooperative behavior using the Public Goods game.
Simulating AI that is forced to cooperate (“Policy 1”) had little effect on increasing human cooperation, as the underlying dilemma remains unchanged.
When given control, humans exploited AI agents to promote defecting behavior (“Policy 2”), showing a tendency to offload cooperative costs onto AI while benefiting from group advantages.
Mimicking AI that mirrors human behavior (“Policy 3”) successfully lowered the threshold for cooperation, demonstrating that responsive AI can encourage more collective effort.
The study warns that “nice” AI alone may not address free-riding behavior, and humans might use AI systems to protect exploitative behavior instead of promoting genuine cooperation.
Researchers emphasize the study’s theoretical framework and highlight limitations, making it non-extrapolatable to real-world settings directly.
The mimicking AI approach suggests a design alternative where AI responds to human actions rather than unconditionally absorbing behavior, potentially benefiting cooperation in large-scale human-AI interactions.