Research: Quantifying and Addressing Algorithmic Bias in AI-Driven Recruitment Pipeline...

Research question: How can bias in AI-assisted hiring funnels be systematically measured and effectively mitigated to ensure fair candidate selection across demographic groups?

Background & Motivation: The increasing adoption of AI tools in recruitment processes offers efficiency but raises concerns about perpetuating or amplifying biases against certain demographic groups. Recent studies have shown that AI models trained on historical hiring data may inherit or even exacerbate systemic biases present in organizations.

Research Gap: While prior work has identified the existence of bias in AI hiring tools, there is a lack of comprehensive frameworks for both reliably measuring bias at multiple stages of AI-assisted hiring funnels and evaluating the effectiveness of mitigation strategies in real-world settings.

Approach & Expected Contribution: This project will analyze datasets from simulated or anonymized real-world hiring funnels processed by AI screening tools. It will apply statistical fairness metrics (e.g., disparate impact, equal opportunity difference) and test mitigation techniques such as pre-processing reweighting and in-processing algorithmic adjustments. The study aims to provide a rigorous assessment of measurement approaches and practical mitigation effectiveness.

Why It Matters: As organizations move toward data-driven talent acquisition, ensuring AI hiring tools are both effective and equitable is critical for compliance, reputation, and diversity goals. This research can inform best practices for practitioners and policymakers in implementing fair AI systems in HR.