Prompts differ but overlap across games and human populations (high, middle, low income; student vs non students) in intuitive ways. They can be used to categorize/compare strategic scenarios and human populations, and potentially gain insight into motivations behind different behaviors.
How well can LLMs mimic human behavior in playing games? What is revealed by the prompts needed to generate different behaviors? They reveal quite a lot:
www.pnas.org/doi/10.1073/...
just published open access with Yutong Xie, Qiaozhu Mei, Walter Yuan
Time to remind ourselves of some observations about how trade appears to help stabilize alliances and prevent international conflict www.gsb.stanford.edu/insights/mat...
