CADEMAS – A Framework for Cooperative Automated Decision-Making Systems

The growing ubiquity of data and artificial intelligence has made it increasingly easy to construct decision models that automate or assist human decision-making. As organizations and infrastructures rely on multiple, heterogeneous Automated Decision-Making (ADM) systems, a new challenge emerges: how to integrate and reconcile their outputs into coherent, explainable, and context-aware collective decisions. This paper proposes a new conceptual and architectural framework, the Cooperative Automated Decision-Making System (CADEMAS), as a foundation for the study and design of cooperative decision architectures. CADEMAS formalizes the interaction among individual ADM systems through an Input Manager, an Output Manager, and a Context Model, allowing the transformation and aggregation of either homogeneous or heterogeneous decision outputs. CADEMAS generalizes cooperation across potentially heterogeneous decision agents, data types and goals, going beyond ensemble learning schemes, which combine homogeneous models over identical data. We argue that CADEMAS represents a necessary paradigm shift: from building isolated decision models to engineering cooperative ecosystems of decision-making entities. This paper outlines the theoretical basis, architectural elements, and implications for future research.