Agent-Based Models as Digital Twins in Management Science

Abstract

Abstract Leveraging advancements in data science, computational capabilities, and artificial intelligence, researchers are constructing so-called digital twins of complex systems as diverse as space vehicles, cars, supply chains, individual patients, power plants, and cities. These digital models, characterized by their high fidelity and grounded in large amounts of frequently updated empirical data, serve as potent instruments for predicting the dynamic behaviours of these complex systems. What is more, they facilitate experimentation with counterfactual scenarios, allowing to compare digital twins with counterfactual siblings. These experiments enable modellers to address research problems that would otherwise remain unanswered due to the unavailability of empirical data, often for practical or ethical considerations. This chapter scrutinizes the challenges associated with crafting digital twins of complex organizations, contending that both technical and ethical hurdles demand careful consideration. While digital twins serve as tools to enhance organizational performance for all stakeholders, they also harbour the potential to be leveraged in ways that may run counter to the interests of certain members within the organization. This chapter posits that the evaluation and potential resolution of these challenges present formidable research opportunities for organizational modellers, underscoring these obstacles as not only impediments but also as intriguing research problems to be addressed.