A Tale of Two Identities: An Ethical Audit of Human and AI-Crafted Personas

TL;DR

This paper critically examines how AI-generated personas, especially those representing minority identities, can perpetuate stereotypes and harm, highlighting the need for better evaluation and validation methods.

Contribution

It provides an ethical audit of LLM-generated personas focusing on racial identity, revealing patterns of representational harm and proposing new evaluation and validation protocols.

Findings

LLMs disproportionately emphasize racial markers

Generated personas often contain culturally coded language

Narratives tend to be syntactically elaborate but reductive

Abstract

As LLMs (large language models) are increasingly used to generate synthetic personas particularly in data-limited domains such as health, privacy, and HCI, it becomes necessary to understand how these narratives represent identity, especially that of minority communities. In this paper, we audit synthetic personas generated by 3 LLMs (GPT4o, Gemini 1.5 Pro, Deepseek 2.5) through the lens of representational harm, focusing specifically on racial identity. Using a mixed methods approach combining close reading, lexical analysis, and a parameterized creativity framework, we compare 1512 LLM generated personas to human-authored responses. Our findings reveal that LLMs disproportionately foreground racial markers, overproduce culturally coded language, and construct personas that are syntactically elaborate yet narratively reductive. These patterns result in a range of sociotechnical harms, including stereotyping, exoticism, erasure, and benevolent bias, that are often obfuscated by superficially positive narrations. We formalize this phenomenon as algorithmic othering, where minoritized identities are rendered hypervisible but less authentic. Based on these findings, we offer design recommendations for narrative-aware evaluation metrics and community-centered validation protocols for synthetic identity generation.