Gender Bias in LLM Hiring Decisions: Evidence from a Japanese Context and Evaluation of Mitigation Strategies

2026-06-16

AI Summary

This study presents the first systematic evaluation of gender bias in large language models (LLMs) within a non-Western context, specifically focusing on Japanese hiring scenarios. By constructing 60 counterfactual Japanese-style résumés paired with 12 linguistically grounded gendered name pairs, the authors conducted 43,200 API calls across five leading LLMs under three conditions: baseline, gender-neutral prompting, and privacy filtering. Results reveal a consistent and significant pro-female bias across all models, with surnames serving as the primary gender signal. Gender-neutral instructions at the prompt level proved ineffective in mitigating bias, whereas removing names nearly eliminated it. Furthermore, privacy filtering mechanisms exhibited compatibility issues with content safety protocols, resulting in a 42% refusal rate for GPT-4o.

Abstract

Large language models (LLMs) are increasingly deployed in hiring workflows, yet most research on gender bias in LLM hiring decisions has focused on English-language, Western-format resumes. This study examines whether pro-female gender bias extends to a Japanese corporate context and evaluates two practical mitigation strategies. Using a counterfactual resume design with 60 Japanese rirekisho-format resumes, 12 name pairs selected on linguistically grounded gender-signal criteria, and five state-of-the-art LLMs (Claude Sonnet 4.6, GPT-4o, DeepSeek-V3, Gemini 2.5 Flash, Llama 3.3 70B), we conducted 43,200 API calls across baseline, prompt instruction, and privacy filter conditions. A crossed random-effects linear mixed model confirms a significant pro-female bias across all five models, replicating Western findings in a non-Western context. A prompt-level gender-neutrality instruction produces no meaningful reduction in bias. A name-reliance analysis formally identifies the candidate name as the primary gender channel: removing the name from the prompt reduces the female effect by nearly its full magnitude. An unexpected incompatibility between the privacy filter and GPT-4o's content safety filter, resulting in a 42% refusal rate, highlights a practical deployment challenge for name anonymization in LLM-assisted recruitment pipelines.

Problem

• gender bias
• large language models
• hiring decisions
• Japanese context
• algorithmic fairness

Innovation

• gender bias
• large language models
• counterfactual resume design
• name anonymization
• cross-cultural evaluation