TLDR: We used LLM-powered program synthesis to automatically model and discover differences between human and LLM strategic behavior
TLDR: We used LLM-powered program synthesis to automatically model and discover differences between human and LLM strategic behavior