Saved in:
Bibliographic Details
Main Authors: Khaokaew, Yonchanok, Salim, Flora D., Züfle, Andreas, Xue, Hao, Anderson, Taylor, MacIntyre, C. Raina, Scotch, Matthew, Heslop, David J
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2504.08260
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912332198707200
author Khaokaew, Yonchanok
Salim, Flora D.
Züfle, Andreas
Xue, Hao
Anderson, Taylor
MacIntyre, C. Raina
Scotch, Matthew
Heslop, David J
author_facet Khaokaew, Yonchanok
Salim, Flora D.
Züfle, Andreas
Xue, Hao
Anderson, Taylor
MacIntyre, C. Raina
Scotch, Matthew
Heslop, David J
contents Generative agents have been increasingly used to simulate human behaviour in silico, driven by large language models (LLMs). These simulacra serve as sandboxes for studying human behaviour without compromising privacy or safety. However, it remains unclear whether such agents can truly represent real individuals. This work compares survey data from the Understanding America Study (UAS) on healthcare decision-making with simulated responses from generative agents. Using demographic-based prompt engineering, we create digital twins of survey respondents and analyse how well different LLMs reproduce real-world behaviours. Our findings show that some LLMs fail to reflect realistic decision-making, such as predicting universal vaccine acceptance. However, Llama 3 captures variations across race and Income more accurately but also introduces biases not present in the UAS data. This study highlights the potential of generative agents for behavioural research while underscoring the risks of bias from both LLMs and prompting strategies.
format Preprint
id arxiv_https___arxiv_org_abs_2504_08260
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Evaluating the Bias in LLMs for Surveying Opinion and Decision Making in Healthcare
Khaokaew, Yonchanok
Salim, Flora D.
Züfle, Andreas
Xue, Hao
Anderson, Taylor
MacIntyre, C. Raina
Scotch, Matthew
Heslop, David J
Computation and Language
Generative agents have been increasingly used to simulate human behaviour in silico, driven by large language models (LLMs). These simulacra serve as sandboxes for studying human behaviour without compromising privacy or safety. However, it remains unclear whether such agents can truly represent real individuals. This work compares survey data from the Understanding America Study (UAS) on healthcare decision-making with simulated responses from generative agents. Using demographic-based prompt engineering, we create digital twins of survey respondents and analyse how well different LLMs reproduce real-world behaviours. Our findings show that some LLMs fail to reflect realistic decision-making, such as predicting universal vaccine acceptance. However, Llama 3 captures variations across race and Income more accurately but also introduces biases not present in the UAS data. This study highlights the potential of generative agents for behavioural research while underscoring the risks of bias from both LLMs and prompting strategies.
title Evaluating the Bias in LLMs for Surveying Opinion and Decision Making in Healthcare
topic Computation and Language
url https://arxiv.org/abs/2504.08260