AI & Humans: Making the Relationship Work

Schneier and Sanders argue that we should manage agentic AI using traditional human management principles. They propose a pragmatic framework based on delegation (parallel tasks), iteration (trial and error), and information sharing (horizontal protocols).

Using human management as a design metaphor is remarkably effective for creating high-performing systems because it leverages our intuitions about coordinating complex work. However, anthropomorphising AI is psychologically and operationally dangerous. These human-centric lessons must be reworked for agents to ensure that fluency does not lead to a reduction in the scrutiny needed to maintain system integrity.