Machine Learning in Hiring

Machine learning (ML) in hiring refers to the use of algorithms that learn patterns from historical data to make predictions about candidates. Common applications include predicting candidate job performance from resume features, identifying candidates most likely to accept an offer, scoring assessment responses, and ranking applicants by predicted fit. ML hiring models are trained on historical data — typically records of past applicants and their outcomes (hired/not hired, performance ratings, tenure). The fundamental bias risk is that historical data reflects historical discrimination: if an organization historically hired fewer women for engineering roles, an ML model trained on that data will learn to prefer male candidates. ML models can also discover and exploit proxy variables that correlate with protected characteristics, creating disparate impact without explicitly using demographic information. Regulations like NYC LL144 and the EU AI Act specifically target ML-powered hiring tools, requiring bias audits, transparency, and human oversight.