Algorithmic fairness as a human-technology interaction problem.

Abstract

Algorithms shape high-stakes decisions across society. While promising efficiency, algorithms also raise fairness concerns. This article proposes that algorithmic fairness is best understood as a human-technology interaction problem rather than a purely technical challenge. Algorithms can reproduce human biases, amplify them through feedback loops, or create new forms of unfairness through objectives, proxies, and seemingly neutral variables. Yet they can make decision processes more explicit, disparities more visible, and actively mitigate discrimination. Fairness depends not only on statistical properties but also on how algorithms are designed, used and experienced by those affected by their decisions. This article therefore offers an interdisciplinary perspective that integrates insights from social justice, psychology, computer science, judgment and decision-making, and management.