Large language models reduce well-being differences but keep social patterns
Large Language Models Substantially Compress Well-Being Inequality but Largely Preserve Its Socioeconomic Structure
Artificial Intelligence
Summary
Measuring how happy people are can be tricky, especially when using computer-generated data. The authors found that large language models tend to show less variation in life satisfaction than real survey data. However, these models still capture important social patterns, like how people with lower incomes tend to have more variation in happiness than wealthier groups. This means such models compress differences but preserve where those differences matter most.
What this means in practice
- •For social data analysts: Use LLM-generated data to study relative social patterns in well-being across income and demographic groups despite lower overall variation.
- •For policy modeling teams: Incorporate large language model outputs to simulate synthetic populations that reflect social well-being gradients for scenario analysis.
Authors
Nattavudh Powdthavee
Abstract
Research using large language models (LLMs) to generate synthetic populations has repeatedly shown that model outputs compress the diversity of human experience. This has raised doubts about whether LLM-generated data can capture meaningful differences within populations. We show that such compression does not necessarily erase the social structure of human heterogeneity. Using 93,901 respondents from 66 countries and territories in Wave 7 of the World Values Survey, we ask six LLMs to predict respondents' life satisfaction from demographic, socioeconomic, and attitudinal profiles. All six models substantially understate the overall dispersion of life satisfaction. Yet after normalizing for these differences in scale, they largely reproduce the human income gradient in well-being inequality: lower-income groups remain relatively more heterogeneous than higher-income groups. The pattern is robust to country fixed effects, equal-country weighting, WVS survey weights, and observed demographic composition, and it extends directionally to employment, education, and perceived control. Fidelity is weaker for extreme outcomes and country-specific gradients. These results show that the amount of heterogeneity preserved by an LLM and the way that heterogeneity is distributed across social groups are distinct properties. LLM-generated populations can therefore substantially compress human variation while retaining meaningful information about where that variation is concentrated.