Algorithmic decision-making in organizations: a systematic review toward an integrated tension alignment framework

What the paper says

Purpose Organizations increasingly employ algorithms to support or automate decision-making processes across various functions, such as personnel performance evaluation, resource allocation and strategic planning. Despite growing research interest, the literature on algorithmic decision-making (ADM) remains fragmented across disciplines and reports mixed findings on its effectiveness. This systematic review aims to synthesize existing literature to develop an integrated understanding of ADM in organizations. Design/methodology/approach Following PRISMA guidelines, this systematic review examined 90 articles to identify patterns in theoretical foundations, antecedents, outcomes, mediating mechanisms and moderators of ADM in organizations. Findings Four theoretical strands – organizational theory and management theories, trust in algorithms, justice and ethics theories, and decision theory and computational rationality – formed the theoretical foundation of ADM research. ADM’s individual and organizational outcomes are shaped by algorithm, task, human and organizational characteristics. These relationships are mediated by perceived fairness, trust in algorithms, role ambiguity and conflict, interpretive labor, and reductionism, and moderated by human-algorithm interaction and the regulatory environment. To reconcile the mixed findings and paradoxes, the Integrative Tension Alignment (ITA) Framework is proposed to conceptualize ADM outcomes as emergent properties of four tensions: transparency ↔ opacity, autonomy ↔ control, human ↔ algorithmic agency, and exploration ↔ exploitation. Originality/value The ITA framework advances ADM scholarship by offering an integrative model that serves as a theoretical foundation for future longitudinal and multi-method empirical testing.