SPAGBias: Uncovering and Tracing Structured Spatial Gender Bias in Large Language Models

2026-04-16

AI Summary

This study addresses the risk that large language models may reproduce or amplify gender biases embedded in spatial organization, particularly in applications like urban planning, where systematic evaluation frameworks are lacking. Introducing gendered space theory into bias research for the first time, this work proposes the first evaluation framework targeting gender bias across 62 categories of urban microspaces. It develops a three-tier diagnostic mechanism—explicit, probabilistic, and constructivist—supported by techniques including forced-choice resampling, token-level asymmetry analysis, and semantic–narrative role parsing. The findings reveal that gender–space associations in models significantly deviate from real-world distributions, manifesting in story generation as coordinated biases in emotion, diction, and social roles. These biases lead to concrete failures in both normative and descriptive tasks, uncovering how gender bias is embedded and reinforced throughout the model training pipeline.

Technology Category

• Natural Language Processing: Ethics — Bias, Fairness, Transparency & Privacy
• Computer Vision: Bias, Fairness & Privacy
• Machine Learning: Ethics, Bias, and Fairness

Application Category

• Social Networks and Social Media: Fairness and bias in social network and social media analysis
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics

Abstract

Large language models (LLMs) are being increasingly used in urban planning, but since gendered space theory highlights how gender hierarchies are embedded in spatial organization, there is concern that LLMs may reproduce or amplify such biases. We introduce SPAGBias - the first systematic framework to evaluate spatial gender bias in LLMs. It combines a taxonomy of 62 urban micro-spaces, a prompt library, and three diagnostic layers: explicit (forced-choice resampling), probabilistic (token-level asymmetry), and constructional (semantic and narrative role analysis). Testing six representative models, we identify structured gender-space associations that go beyond the public-private divide, forming nuanced micro-level mappings. Story generation reveals how emotion, wording, and social roles jointly shape "spatial gender narratives". We also examine how prompt design, temperature, and model scale influence bias expression. Tracing experiments indicate that these patterns are embedded and reinforced across the model pipeline (pre-training, instruction tuning, and reward modeling), with model associations found to substantially exceed real-world distributions. Downstream experiments further reveal that such biases produce concrete failures in both normative and descriptive application settings. This work connects sociological theory with computational analysis, extending bias research into the spatial domain and uncovering how LLMs encode social gender cognition through language.

Problem

• spatial gender bias
• large language models
• gendered space
• urban planning
• social bias

Innovation

• spatial gender bias
• large language models
• bias evaluation framework
• micro-space taxonomy
• bias tracing