Robust, Scalable Hybrid Decision Networks

摘要

We present a framework for designing and evaluating systemically-robust and efficient hybrid (human and machine) decision networks. We propose the degree of diversity for enterprise variables be chosen according to the expected demands of the decision scenario, with margins for risk and uncertainty. We illustrate the use of the model by comparing two mission case studies and outline a theoretical approach for analyzing enterprise settings. As the robustness of the enterprise framework ultimately rests on the ability to encode and adapt expertise in software, we explain our approaches to the development of robust intelligence.