Large Model Consumer Behavior Theory: When AI Becomes the Subject of Consumption Decisions

This article discusses the transformation of large language models (LLMs) from auxiliary tools to autonomous agents participating in consumption decisions, pointing out that this poses fundamental challenges to traditional consumer behavior theory, and proposes the basic framework and open questions of the emerging research field of "Large Model Consumer Behavior Theory". Original source: arXiv platform, author team (arXiv:2606.18005v1), release date: June 16, 2026, link: http://arxiv.org/abs/2606.18005v1.

Traditional consumer behavior theories (such as the rational man hypothesis and cognitive biases in behavioral economics) all take humans as the decision-making subject. However, when LLMs act as autonomous agents to make decisions for users, questions such as how AI understands human preferences, whether decision-making patterns follow human logic, and changes in market demand aggregation mechanisms are all beyond the explanatory scope of traditional theories.

The core of this theory is to expand the analysis object to AI agents in the agent market and the users behind them, focusing on three key issues: 1. Mapping and agency of human preferences (information transformation, bias, gap between real intentions); 2. Agent decision-making mechanism (similarities and differences between statistical rationality and human rationality, machine bias); 3. Market aggregation effect (formation of demand curves, impact of homogenization, changes in supply-demand balance).

The new theory integrates scattered research such as LLM decision-making capabilities, human behavior simulation, and preference induction, and reveals internal connections through an economic lens. It also reconsiders traditional assumptions: the rationality hypothesis (statistical rationality of AI vs. rationality/irrationality of humans), and the heterogeneity hypothesis (similarity of AI agents vs. natural differences of humans).

This field faces several open questions: 1. Alignment issue (consistency between AI decisions and users' long-term interests, effectiveness of RLHF technology); 2. Preference representation (representation and update of preference structures dependent on fuzzy and dynamic contexts); 3. Market dynamics (changes in competitive landscape, adjustment of supplier strategies, risks of market failure such as algorithmic collusion).

This theory has interdisciplinary characteristics and needs to combine perspectives from classical/behavioral economics, natural language processing, ethics, law, sociology, and other disciplines. Interdisciplinary integration is both a challenge (method adaptation and fusion) and an opportunity (knowledge innovation).

Practical significance: Platform designers can optimize recommendation algorithms and interfaces; regulators need to grasp the laws of the agent market to formulate policies; consumers can use AI tools more wisely. Ethical and legal issues: The attribution of responsibility for AI decisions needs to be clarified and resolved based on theoretical analysis.

The Large Model Consumer Behavior Theory marks the intersection of economics and AI research, expands the boundaries of consumption theory, and provides tools for understanding consumption patterns in the AI era. Although it is in its infancy, its future impact will be far-reaching, and it will become an important topic of common concern for academia and industry.