The Personality Trap: How LLMs Embed Bias When Generating Human-Like Personas

2026-02-03

AI Summary

This study reveals significant sociodemographic and psychological trait biases in large language models (LLMs) when generating anthropomorphized characters, particularly manifesting as stereotyping and potential pathologization of marginalized groups such as LGBTQ+ individuals. Leveraging prompt engineering grounded in established personality psychology scales, the authors generated synthetic populations using five mainstream LLMs and systematically evaluated the alignment between assigned sociodemographic attributes and target personality traits, while quantifying WEIRD (Western, Educated, Industrialized, Rich, Democratic) bias. The research uncovers, for the first time, a consistent and pronounced model preference for young, highly educated, Western, White, heterosexual individuals across all tested LLMs. Notably, under high Psychoticism conditions, certain models erroneously associated non-binary or LGBTQ+ identities with pathological personality traits, exposing latent discriminatory patterns and serious ethical concerns in AI-generated personas.

Technology Category

• Humans and AI: Learning Human Values and Preferences
• Natural Language Processing: Ethics — Bias, Fairness, Transparency & Privacy
• Philosophy and Ethics of AI: Bias, Fairness & Equity

Application Category

• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Social Networks and Social Media: Generative AI / large language models and their impact on social systems
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems

Abstract

This paper examines biases in large language models (LLMs) when generating synthetic populations from responses to personality questionnaires. Using five LLMs, we first assess the representativeness and potential biases in the sociodemographic attributes of the generated personas, as well as their alignment with the intended personality traits. While LLMs successfully reproduce known correlations between personality and sociodemographic variables, all models exhibit pronounced WEIRD (western, educated, industrialized, rich and democratic) biases, favoring young, educated, white, heterosexual, Western individuals with centrist or progressive political views and secular or Christian beliefs. In a second analysis, we manipulate input traits to maximize Neuroticism and Psychoticism scores. Notably, when Psychoticism is maximized, several models produce an overrepresentation of non-binary and LGBTQ+ identities, raising concerns about stereotyping and the potential pathologization of marginalized groups. Our findings highlight both the potential and the risks of using LLMs to generate psychologically grounded synthetic populations.

Problem

• bias
• large language models
• synthetic populations
• personality traits
• WEIRD

Innovation

• LLM bias
• synthetic personas
• WEIRD bias
• personality traits
• marginalized groups