# Robustly Improving LLM Fairness in Realistic Settings via Interpretability

arXiv: https://arxiv.org/abs/2506.10922 · HTML: https://mlsanity.com/paper/2506.10922/

## Key facts

- Cited by 18 papers in the map (data through 14 Oct 2026).
- Builds on 29 of its 43 references that are in the map.
- First posted on arXiv 12 Jun 2025.

## Builds on (most cited first, 29 of 29)

- [Measuring Massive Multitask Language Understanding](https://mlsanity.com/paper/2009.03300/) (2020)
- [Representation Engineering: A Top-Down Approach to AI Transparency](https://mlsanity.com/paper/2310.01405/) (2023)
- [Man is to Computer Programmer as Woman is to Homemaker? Debiasing Word Embeddings](https://mlsanity.com/paper/1607.06520/) (2016)
- [Inference-Time Intervention: Eliciting Truthful Answers from a Language Model](https://mlsanity.com/paper/2306.03341/) (2023)
- [Steering Llama 2 via Contrastive Activation Addition](https://mlsanity.com/paper/2312.06681/) (2023)
- [Refusal in Language Models Is Mediated by a Single Direction](https://mlsanity.com/paper/2406.11717/) (2024)
- [Steering Language Models With Activation Engineering](https://mlsanity.com/paper/2308.10248/) (2023)
- [Semantics derived automatically from language corpora contain human-like biases](https://mlsanity.com/paper/1608.07187/) (2016)
- [The Geometry of Truth: Emergent Linear Structure in Large Language Model Representations of True/False Datasets](https://mlsanity.com/paper/2310.06824/) (2023)
- [CrowS-Pairs: A Challenge Dataset for Measuring Social Biases in Masked Language Models](https://mlsanity.com/paper/2010.00133/) (2020)
- [Null It Out: Guarding Protected Attributes by Iterative Nullspace Projection](https://mlsanity.com/paper/2004.07667/) (2020)
- [Monitoring Reasoning Models for Misbehavior and the Risks of Promoting Obfuscation](https://mlsanity.com/paper/2503.11926/) (2025)
- [Persistent Anti-Muslim Bias in Large Language Models](https://mlsanity.com/paper/2101.05783/) (2021)
- [LEACE: Perfect linear concept erasure in closed form](https://mlsanity.com/paper/2306.03819/) (2023)
- [Finding Alignments Between Interpretable Causal Variables and Distributed Neural Representations](https://mlsanity.com/paper/2303.02536/) (2023)
- [Causal Mediation Analysis for Interpreting Neural NLP: The Case of Gender Bias](https://mlsanity.com/paper/2004.12265/) (2020)
- [Beyond Memorization: Violating Privacy Via Inference with Large Language Models](https://mlsanity.com/paper/2310.07298/) (2023)
- [Learning to Generate Reviews and Discovering Sentiment](https://mlsanity.com/paper/1704.01444/) (2017)
- [Evaluating and Mitigating Discrimination in Language Model Decisions](https://mlsanity.com/paper/2312.03689/) (2023)
- [Identifying and Controlling Important Neurons in Neural Machine Translation](https://mlsanity.com/paper/1811.01157/) (2018)
- [Mitigating Language-Dependent Ethnic Bias in BERT](https://mlsanity.com/paper/2109.05704/) (2021)
- [Are DeepSeek R1 And Other Reasoning Models More Faithful?](https://mlsanity.com/paper/2501.08156/) (2025)
- [Designing a Dashboard for Transparency and Control of Conversational AI](https://mlsanity.com/paper/2406.07882/) (2024)
- [Refusal in LLMs is an Affine Function](https://mlsanity.com/paper/2411.09003/) (2024)
- [JobFair: A Framework for Benchmarking Gender Hiring Bias in Large Language Models](https://mlsanity.com/paper/2406.15484/) (2024)
- [Evaluating Bias in LLMs for Job-Resume Matching: Gender, Race, and Education](https://mlsanity.com/paper/2503.19182/) (2025)
- [FLEX: A Benchmark for Evaluating Robustness of Fairness in Large Language Models](https://mlsanity.com/paper/2503.19540/) (2025)
- [On the Effectiveness and Generalization of Race Representations for Debiasing High-Stakes Decisions](https://mlsanity.com/paper/2504.06303/) (2025)
- [The Impact of Disability Disclosure on Fairness and Bias in LLM-Driven Candidate Selection](https://mlsanity.com/paper/2506.00256/) (2025)

## Built on by (most cited first, 18 of 18)

- [When Chain of Thought is Necessary, Language Models Struggle to Evade Monitors](https://mlsanity.com/paper/2507.05246/) (2025)
- [Population-Aligned Persona Generation for LLM-based Social Simulation](https://mlsanity.com/paper/2509.10127/) (2025)
- [Measuring Chain-of-Thought Monitorability Through Faithfulness and Verbosity](https://mlsanity.com/paper/2510.27378/) (2025)
- [Thought Branches: Interpreting LLM Reasoning Requires Resampling](https://mlsanity.com/paper/2510.27484/) (2025)
- [Mechanistic Indicators of Understanding in Large Language Models](https://mlsanity.com/paper/2507.08017/) (2025)
- [Biases in the Blind Spot: Detecting What LLMs Fail to Mention](https://mlsanity.com/paper/2602.10117/) (2026)
- [Do Bias Benchmarks Generalise? Evidence from Voice-based Evaluation of Gender Bias in SpeechLLMs](https://mlsanity.com/paper/2510.01254/) (2025)
- [C2-Faith: Benchmarking LLM Judges for Causal and Coverage Faithfulness in Chain-of-Thought Reasoning](https://mlsanity.com/paper/2603.05167/) (2026)
- [Ads in AI Chatbots? An Analysis of How Large Language Models Navigate Conflicts of Interest](https://mlsanity.com/paper/2604.08525/) (2026)
- [AgentFairBench: Do LLM Agents Discriminate When They Act?](https://mlsanity.com/paper/2606.16723/) (2026)
- [Value Leakage: An LLM's Answers Are Silently Shaped by Its Own Values](https://mlsanity.com/paper/2607.14345/) (2026)
- [Can SAEs reveal and mitigate racial biases of LLMs in healthcare?](https://mlsanity.com/paper/2511.00177/) (2025)
- [Towards Ethical Multi-Agent Systems of Large Language Models: A Mechanistic Interpretability Perspective](https://mlsanity.com/paper/2512.04691/) (2025)
- [Identifying and Mitigating Gender Cues in Academic Recommendation Letters: An Interpretability Case Study](https://mlsanity.com/paper/2604.12337/) (2026)
- [Gender Bias in LLM Hiring Decisions: Evidence from a Japanese Context and Evaluation of Mitigation Strategies](https://mlsanity.com/paper/2606.18649/) (2026)
- [Who Do Language Models Think Is Competent? A Mechanistic Analysis of Occupational Bias](https://mlsanity.com/paper/2608.20347/) (2026)
- [Do LLMs Choose Like Humans? Using Cognitive Theory to Evaluate LLM Decision-Making](https://mlsanity.com/paper/2609.22225/) (2026)
- [Does Thinking Help Fairness? Reasoning Tokens Resolve Some Biases but Create More](https://mlsanity.com/paper/2609.30768/) (2026)

## Abstract (arXiv)

Large language models (LLMs) are increasingly deployed in high-stakes hiring applications, making decisions that directly impact people's careers and livelihoods. While prior studies suggest simple anti-bias prompts can eliminate demographic biases in controlled evaluations, we find these mitigations fail when realistic contextual details are introduced. We address these failures through internal bias mitigation: by identifying and neutralizing sensitive attribute directions within model activations, we achieve robust bias reduction across all tested scenarios. Across leading commercial (GPT-4o, Claude 4 Sonnet, Gemini 2.5 Flash) and open-source models (Gemma-2 27B, Gemma-3, Mistral-24B), we find that adding realistic context such as company names, culture descriptions from public careers pages, and selective hiring constraints (e.g.,``only accept candidates in the top 10%") induces significant racial and gender biases (up to 12% differences in interview rates). When these biases emerge, they consistently favor Black over White candidates and female over male candidates across all tested models and scenarios. Moreover, models can infer demographics and become biased from subtle cues like college affiliations, with these biases remaining invisible even when inspecting the model's chain-of-thought reasoning. To address these limitations, our internal bias mitigation identifies race and gender-correlated directions and applies affine concept editing at inference time. Despite using directions from a simple synthetic dataset, the intervention generalizes robustly, consistently reducing bias to very low levels (typically under 1%, always below 2.5%) while largely maintaining model performance. Our findings suggest that practitioners deploying LLMs for hiring should adopt more realistic evaluation methodologies and consider internal mitigation strategies for equitable outcomes.