Rethinking Corporate Knowledge for Human-AI Decision-Making

Corporate knowledge is often scattered across various systems, expert insights, and documents, primarily structured for human understanding. However, with the increasing deployment of AI systems in decision-making roles, there's a growing need to make this knowledge accessible and usable for AI. This raises fundamental questions about how organizations should store, maintain, and present their knowledge base to serve both human and artificial intelligence effectively. This paper addresses two core challenges: first, optimizing knowledge management for dual human-AI access, and second, strategically allocating decision-making agency between humans and AI, considering varying task risks and uncertainties. It argues for a re-evaluation of current knowledge infrastructure to support AI's informational needs. The authors introduce a framework that correlates specific task attributes and the availability of relevant knowledge with appropriate agency allocations and control mechanisms. This framework is illustrated through examples in manufacturing, such as routine quality inspection and strategic factory location decisions, providing a practical guide for rethinking human-AI collaboration in corporate environments.