Counterfactual, Per-Decision Bias Auditing for Automated Hiring: Localizing and Explaining Disparate Impact in Applicant Tracking Systems

WHEN FAIR AI BECOMES UNFAIR: A COUNTERFACTUAL AUDIT OF POSITIONAL BIAS IN LARGE LANGUAGE MODELS FOR HIRING DECISIONS

Findings indicate that state-of-the-art LLMs can achieve a high degree of demographic neutrality; fundamental artefacts such as positional bias can nonetheless produce severely discriminatory outcomes; and bias auditing must extend beyond demographic parity to interaction artefacts and ecosystem structure.

AI and Automated Decision-making Systems in Employment: A Look at State General Statutes Addressing of Bias Audits, Transparency, and Legal Accountability in the US

Increasingly, companies are utilizing electronic decision-making systems powered by AI for the hiring, evaluation, and upskilling of employees. While automation may help, evidence suggests that algorithmic devices can replicate and potentially intensify social bias. This research examines whether the use of AI in hiring and selection is legal and fair, by creating discriminatory processes into employment practices. This study examines bias audits, transparency obligations, and new regulations such as New York City’s Local Law 144 in a doctrinal legal process. This research examines whether the proposed Bias Notification Duty will enhance accountability. The study indicates that while AI facilitates efficient processes in the workplace, it also risks increasing discrimination and bias. It is known that humans design AI and they might have inherent biases. These computer professionals then create technological systems like AI that inherit these biases. HR managers and leaders have obligations for the responsible implementation of algorithmic systems to promote fairness and equality in HR processes. The findings in examining legislation addressing AI for practicing managers indicate that the integration of AI into employment decisions carries profound operational, legal, and ethical implications. As managers make greater use of automated decision-making systems, they do not avoid accountability; rather, they increase it.

Applied and Filtered: An End-to-End Algorithmic Fairness Audit of A Public Employment Agency

Algorithmic fairness evaluation commonly assesses AI systems as bounded technical components, abstracting away the organizational context in which they operate. We present, to our knowledge, the first independent end-to-end fairness audit of a semi-automated hiring system operated by Barcelona Activa, a public employment agency using the third-party TalentClue platform for candidate search and shortlisting. We analyze approximately 497,000 candidate-vacancy pipeline entries from September 2017 to September 2022, covering seven pipeline stages that span automated processing, human discretion, candidate data, and employer decisions. Aggregate outcomes across binary genders are statistically indistinguishable, yet this parity masks substantial disparities by salary level, age, and gender identity. Women face adverse impact in mid-salary shortlisting (DIR = 0.786, p<0.001), alongside salary disparities in 15 of 20 sectors and a compounded disadvantage for women aged 46-55 (DIR = 0.77). Non-binary candidates are shortlisted at less than one third the rate of men (DIR = 0.295), although this estimate rests on a small sample (N = 285). Candidates aged 55 and over are entirely absent from the pipeline despite comprising 15.6% of Barcelona's labor force. The gender gap in shortlisting narrows over time, from 6.5 percentage points in 2017 to 1.3 in 2022. The audit further reveals a vendor-deployer information asymmetry: Barcelona Activa lacks access to key information about TalentClue's matching logic and evaluation. Fairness outcomes can thus arise from interactions among automated processing, human discretion, data quality, vendor opacity, and pipeline structure. We build on prior calls for sociotechnical, end-to-end fairness evaluation, showing empirically why model-level assessment alone can be insufficient for understanding fairness in deployed systems.

White-Hat Testing for the Ballot Box: A Framework for Election AI Auditing

Counterfactual Bias Testing for Application Tracking System

Automated candidate-job matching systems are increasingly classified as high-risk AI under emerging regulation, yet auditing them for demographic bias is expensive: classical correspondence-audit studies require hand-crafted resumes and manual submission, which does not scale to fast pipeline retraining cycles. This paper presents a general, reusable methodology that (1) uses task-specialized LLM agents to synthesize identity-neutral base resumes and inject controlled demographic treatments across five protected-characteristic axes (sex/gender, age, residence, language, disability), producing a K x (1+N) correspondence-audit matrix; (2) qualitatively flags inferred protected characteristics per an EU AI Act-aligned prompt; (3) ranks candidates against a job description via a fine-tuned sentence-embedding model and cosine similarity; and (4) computes a nine-metric fairness suite spanning counterfactual (score delta, mean absolute rank change, flip rate), group-fairness (top-K retention, four-fifths/impact ratio), and merit-aware (Recall@K, nDCG@K, equal opportunity, equalized odds) families, each with bootstrap confidence intervals, significance tests, and Benjamini-Hochberg correction, culminating in an automated PASS/INVESTIGATE/FAIL report with a composite risk score. On an example corpus of 5 job orders, 100 base candidates, and 10 demographic treatments (90 metric x variant evaluations): score shifts, top-K retention, and merit-aware rate gaps stay within tolerance for every treatment, but a rank-stability metric (MARC) and nDCG@K each surface borderline findings - including one on the neutral baseline itself - that a score- or retention-only view would miss. The results argue for multi-metric, multi-family auditing over any single aggregate score, and for LLM-agent-generated audits as a practical, low-cost complement to human-curated audits for any candidate-job matching pipeline.

Fairness trade-offs in expert evaluation: information, judgment, and procedural design

Evaluation systems often seek to reduce bias, prevent manipulation, preserve relevant information, maintain comparability, and treat candidates equally, but these fairness goals can conflict. This paper develops a framework for understanding such trade-offs in expert evaluation—including peer review, grant panels, academic hiring, and artistic competitions—where quality is multidimensional and legitimate disagreement among evaluators is possible. We treat the choice among evaluation procedures as a problem of constitutional design: rules must be selected before the specific candidates, evaluators, and circumstances that will test them are known. Classical music competitions are our primary analytical setting because their rules, scores, and controversies make these trade-offs unusually visible, but the framework applies to other settings in which expert judgments are aggregated under incomplete information. The framework highlights two core tensions: information control and outlier rules. First, when information is both quality-relevant and identity-revealing, procedures that suppress bias-relevant cues may also remove information with genuine evaluative content. Second, because sincere minority judgment and strategic manipulation can generate similar observed score patterns, score-based exclusion rules may reduce manipulation while limiting legitimate dissent. Two further trade-offs concerning transparency and standardization are developed as illustrations. The analysis shows that no procedure can eliminate all vulnerabilities at once. Beyond this design implication, the framework also helps explain why evaluation rules differ across stages and settings, and why well-intentioned reforms may create new vulnerabilities.