LLMs show significant pro-female bias on Japanese resumes across five models; name removal nearly eliminates the effect while prompt instructions do not, and privacy filters trigger high refusal rates on GPT-4o.
Robustly improving llm fairness in realistic settings via interpretability
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
years
2026 2representative citing papers
citing papers explorer
-
Gender Bias in LLM Hiring Decisions: Evidence from a Japanese Context and Evaluation of Mitigation Strategies
LLMs show significant pro-female bias on Japanese resumes across five models; name removal nearly eliminates the effect while prompt instructions do not, and privacy filters trigger high refusal rates on GPT-4o.
- Ads in AI Chatbots? An Analysis of How Large Language Models Navigate Conflicts of Interest