Advances in artificial intelligence are increasingly leading to the automation and augmentation of decision processes in work contexts. Although research originally generally focused upon decision-makers, the perspective of those targeted by automated or augmented decisions (whom we call “second parties”) and parties who observe the effects of such decisions (whom we call “third parties”) is now growing in importance and attention. We review the expanding literature investigating reactions to automated and augmented decision-making by second and third parties. Specifically, we explore attitude (e.g., evaluations of trustworthiness), perception (e.g., fairness perceptions), and behavior (e.g., reverse engineering of automated decision processes) outcomes of second and third parties. Additionally, we explore how characteristics of the a) decision-making process, b) system, c) second and third party, d) task, and e) outputs and outcomes moderate these effects, and provide recommendation for future research. Our review summarizes the state of the literature in these domains, concluding a) that reactions to automated decisions differ across situations in which there is remaining human decision control (i.e., augmentation contexts), b) that system design choices (e.g., transparency) are important but underresearched, and c) that the generalizability of findings might suffer from excessive reliance on specific research methodologies (e.g., vignette studies).

人工智能的进步越来越多地导致工作环境中决策过程的自动化和增强。虽然最初的研究通常集中在决策者身上，但自动化或增强决策所针对的人（我们称之为“第二方”）和观察此类决策影响的各方（我们称之为“第三方”）的观点现在正在增长在重要性和关注度上。我们回顾了调查对第二方和第三方自动化和增强决策的反应的不断扩大的文献。具体而言，我们探讨了第二方和第三方的态度（例如，可信度评估）、感知（例如，公平感知）和行为（例如，自动决策过程的逆向工程）结果。此外，我们探讨了 a) 决策过程、b) 系统、c) 第二方和第三方、d) 任务和 e) 输出和结果的特征如何调节这些影响，并为未来的研究提供建议。我们的综述总结了这些领域的文献状况，得出结论：a) 对自动化决策的反应因存在剩余人类决策控制（即增强上下文）的情况而异，b）系统设计选择（例如，透明度）是重要但研究不足，以及 c) 研究结果的普遍性可能会因过度依赖特定的研究方法（例如，小插图研究）而受到影响。