Investigating social alignment via mirroring in a system of interacting language models
This work addresses the scalability limitations in sociology for studying social alignment, though it is incremental as it applies a computational framework to a known mechanism.
The authors tackled the problem of understanding how mirroring behavior affects social alignment in multi-agent systems by simulating interacting large language models, finding that system behavior is strongly influenced by communication range and exacerbated by higher mirroring rates.
Alignment is a social phenomenon wherein individuals share a common goal or perspective. Mirroring, or mimicking the behaviors and opinions of another individual, is one mechanism by which individuals can become aligned. Large scale investigations of the effect of mirroring on alignment have been limited due to the scalability of traditional experimental designs in sociology. In this paper, we introduce a simple computational framework that enables studying the effect of mirroring behavior on alignment in multi-agent systems. We simulate systems of interacting large language models in this framework and characterize overall system behavior and alignment with quantitative measures of agent dynamics. We find that system behavior is strongly influenced by the range of communication of each agent and that these effects are exacerbated by increased rates of mirroring. We discuss the observed simulated system behavior in the context of known human social dynamics.