New Theory Explores LLM Consumer Behavior in Agentic Markets

The increasing deployment of large language models (LLMs) as autonomous agents making consumption decisions for users necessitates a new theoretical framework. Traditional consumer theory primarily models human decision-makers, but the rise of agentic markets introduces fundamental questions about how preferences are translated and acted upon by AI. This paper lays the groundwork for "LLM Consumer Behavior Theory," a novel field of study. It integrates concepts from classical and behavioral economics with advancements in Natural Language Processing to formalize how human preferences are represented and executed by LLM-based agents. Furthermore, it examines how these agent-level decisions collectively shape market demand. The theory aims to unify existing, disparate research on LLM decision-making, human behavior simulation, and preference elicitation through a common economic perspective. It highlights areas where traditional economic assumptions, such as rationality and heterogeneity, may not hold true in markets driven by AI agents. While not providing empirical validation, the paper defines the scope of this new field and identifies critical open research questions concerning alignment, preference representation, and the dynamics of agentic markets.