Large language models that replace human participants can harmfully misportray and flatten identity groups

TL;DR

This paper critically examines the limitations of large language models in accurately representing social identities, highlighting potential harms when used as replacements for human participants in social science research.

Contribution

It analytically and empirically demonstrates that current LLMs tend to misportray and flatten demographic identities, raising concerns about their use in socially sensitive applications.

Findings

LLMs often misrepresent demographic groups

Empirical evidence from 3200 participants across 16 identities

Inference techniques can reduce, but not eliminate, harms

Abstract

Large language models (LLMs) are increasing in capability and popularity, propelling their application in new domains -- including as replacements for human participants in computational social science, user testing, annotation tasks, and more. In many settings, researchers seek to distribute their surveys to a sample of participants that are representative of the underlying human population of interest. This means in order to be a suitable replacement, LLMs will need to be able to capture the influence of positionality (i.e., relevance of social identities like gender and race). However, we show that there are two inherent limitations in the way current LLMs are trained that prevent this. We argue analytically for why LLMs are likely to both misportray and flatten the representations of demographic groups, then empirically show this on 4 LLMs through a series of human studies with 3200 participants across 16 demographic identities. We also discuss a third limitation about how identity prompts can essentialize identities. Throughout, we connect each limitation to a pernicious history of epistemic injustice against the value of lived experiences that explains why replacement is harmful for marginalized demographic groups. Overall, we urge caution in use cases where LLMs are intended to replace human participants whose identities are relevant to the task at hand. At the same time, in cases where the benefits of LLM replacement are determined to outweigh the harms (e.g., the goal is to supplement rather than fully replace, engaging human participants may cause them harm), we provide inference-time techniques that we empirically demonstrate do reduce, but do not remove, these harms.