Observing Micromotives and Macrobehavior of Large Language Models

ACL ARR 2025 February Submission6046 Authors

Abstract: Thomas C. Schelling, awarded the 2005 Nobel Memorial Prize in Economic Sciences, pointed out that "individuals decisions (micromotives), while often personal and localized, can lead to societal outcomes (macrobehavior) that are far more complex and different from what the individuals intended." The current research related to large language models' (LLMs') micromotives, such as preferences or biases, assumes that users will make more appropriate decisions once LLMs are devoid of preferences or biases. However, the NLP community has rarely examined how LLMs might influence society's macrobehavior. In this paper, we follow the design of Schelling's model of segregation to observe the relationship between the micromotives and macrobehavior of LLMs.Our results not only align with current bias evaluation frameworks but also demonstrate our model's capability to effectively simulate how micromotives translate into macrobehavior. Our findings indicate that widespread adoption of LLM suggestions leads to societal segregation, regardless of the LLMs' bias levels. This calls for reconsidering both the mitigation of LLMs' micromotives and their broader societal impact.

Paper Type: Long

Research Area: Computational Social Science and Cultural Analytics

Research Area Keywords: Schelling Model Segeragation

Contribution Types: Model analysis & interpretability

Languages Studied: English

Submission Number: 6046