Building better AI agents for adversarial games costs less than adding reasoning

Original: Context, Reasoning, and Hierarchy: A Cost-Performance Study of Compound LLM Agent Design in an Adversarial POMDP

Source: arxiv.org ↗