How can algorithmic bias manifest in performance reviews and promotions, according to the chapter?

Algorithmic bias can affect performance reviews and promotions when systems trained on historical data or proxy signals repeat existing inequalities. Productivity algorithms may rank employees who work traditional hours more favorably, disadvantaging working parents and employees with flexible requirements, while predictive analytics may label employees as low performers or flight risks based on career breaks or job moves, patterns that occur more often among women and minority employees.

The chapter explains that performance reviews and promotions are not immune to algorithmic bias. Productivity algorithms can prefer employees working during standard, traditional hours, which puts working parents and those relying on flexible scheduling at a disadvantage. Predictive-analytics algorithms can also unfairly classify employees as flight risks or underperformers based on proxy variables such as changing jobs or taking a career break; these patterns more frequently affect female and minority employees. For example, women who return after maternity leave may be mislabeled as attrition risks or underperforming staff. The chapter further notes that these algorithms are often opaque, similar to a black box, so workers and even managers cannot see the reasoning behind an automated rating and have little ability to challenge it. In this way, algorithmic bias can entrench systemic inequalities and weaken employee trust in promotion and review decisions.