People-Centric Operations in Digitized Control Rooms

Despite this shift, humans remain central to operations, as they retain discretionary power and make the final decisions. Consequently, human-algorithm interaction is highly relevant to the future of work. Prior literature has primarily examined this interaction either from the algorithmic perspective or by focusing on the factors that influence humans' decisions to override algorithmic recommendations. However, beyond investigating humans' specific decisions, it is important to regard them as intelligent decision makers with their own judgments, which in turn shape the performance of human-algorithm interactions. Building on this view, a people-centric analysis should be undertaken in the context of the future of work, such as in settings where humans collaborate with algorithmic systems and serve in supervisory roles. Motivated by these aspects, this research aims to answer the following main research question: What people-centric operations (PCO)-related factors influence human operational performance in contexts where individuals work alongside algorithmic decision-making systems, and how can organizations learn from these factors to enhance overall operational performance? This PhD addresses this question through a systematic literature review that focuses on employee-centric operations, as well as three distinct empirical studies. Specifically, the literature review synthesizes existing accomplishments, identifies current gaps, and outlines potential directions for future research. The first empirical study builds upon cognitive load theory to examine how workload affects performance in environments involving algorithm-based decision-making. It further explores whether this effect is moderated by individuals' overriding automation tendency (OAT). The second empirical study builds upon experiential learning theory and employs the knowledge transfer mechanism described by structure-mapping theory to examine how experience, gained through both active and inactive involvement, affects performance. It further differentiates between experience acquired from primary versus non-primary tasks within the active and inactive involvement framework and tests their respective impacts on performance. The third empirical study adopts an exploratory and practice-oriented perspective, grounded in the premise that appropriate overrides are critical to employee performance in automated environments. This study investigates the average treatment effect of overriding algorithmic decisions and further explores whether this effect varies across different task characteristics. In addition, this study further reviews existing methodologies for estimating heterogeneous treatment effects using observational data, thereby laying the foundation for developing a real-time, context-aware decision-support system that can guide humans on when to override algorithmic decisions. It is worth mentioning that all three empirical studies were conducted at the traffic control centers of Infrabel, Belgium's railway infrastructure company. In this setting, traffic controllers manage railway traffic by overseeing algorithmic systems, a characteristic of the future of work that this dissertation focuses on. The real-world data were collected to provide a realistic representation of the setting, enabling us to test hypotheses derived from theoretical arguments and to identify challenges that exist in practice. Overall, this research provides a deeper understanding of how PCO-related factors (i.e., workload, individuals' tendency to override algorithmic systems, and experience) influence operational performance in the context of employees interacting with algorithm-based decision-making systems. It further explores the potential for a decision support system designed to assist employees in determining when it is appropriate to override automated decisions.