arXiv:2607.28347 · v2

LLMs struggle to simulate human belief updates in controlled environments

首次提交 2026-07-30 23:19

版本更新 2026-11-02 20:32

雷达观测 2026-11-04 23:42

摘要

LLMs are increasingly deployed as proxies for human study participants in social science experiments, yet the fidelity of this practice has rarely been tested directly. We test whether six LLMs can simulate individual human belief updates, comparing LLM outputs 1-to-1 against ground truth data from 391 UK participants on Prolific, who updated their stances on three discussion topics after reading Reddit comments. Each participant was simulated by an LLM conditioned on a persona derived from their demographic and personality trait data. Given participants' actual initial stances, five of the six models showed no systematic directional bias in their simulated post-stances, although the size of their individual-level errors varied considerably between models. All six models fail to simulate initial stances themselves and to produce faithful belief updates from self-generated stances. Three systematic biases emerge across all models: overrepresentation of neutral positions, more frequent but smaller belief shifts than humans, and a failure to rank comments by convincingness. Demographic and personality trait personas had no consistent effect on fidelity. We conclude that without grounding in realistic initial stances, LLM simulations of human belief dynamics cannot be considered reliable, a condition that current multi-round social media simulations rarely meet.