CoMPosT: Characterizing and Evaluating Caricature in LLM Simulations
This addresses the issue of perpetuating stereotypes in social science and survey applications, though it is incremental by building on existing concerns.
The paper tackles the problem of evaluating the quality of LLM simulations of human demographics, finding that GPT-4 simulations of certain groups and topics are highly susceptible to caricature, as measured by individuation and exaggeration criteria.
Recent work has aimed to capture nuances of human behavior by using LLMs to simulate responses from particular demographics in settings like social science experiments and public opinion surveys. However, there are currently no established ways to discuss or evaluate the quality of such LLM simulations. Moreover, there is growing concern that these LLM simulations are flattened caricatures of the personas that they aim to simulate, failing to capture the multidimensionality of people and perpetuating stereotypes. To bridge these gaps, we present CoMPosT, a framework to characterize LLM simulations using four dimensions: Context, Model, Persona, and Topic. We use this framework to measure open-ended LLM simulations' susceptibility to caricature, defined via two criteria: individuation and exaggeration. We evaluate the level of caricature in scenarios from existing work on LLM simulations. We find that for GPT-4, simulations of certain demographics (political and marginalized groups) and topics (general, uncontroversial) are highly susceptible to caricature.