Use AI as the production engine, not the design authority

AI governanceCADEDesign authority

CADE required fast production of research, planning artifacts, turn content, and controller materials, but the training effect depended on human judgment about purpose, boundaries, success criteria, and execution risk.

AI would accelerate production while human design authority governed the training problem, source approval, turn structure, adjudication logic, controller discipline, and post-execution changes.

Let AI generate the exercise design directly

Pros
  • Fastest initial output
  • Low manual drafting burden
Cons
  • Weak control over training purpose
  • Higher risk of plausible but unusable exercise logic
  • No defensible boundary between generated content and approved design

Use AI only for editing and formatting

Pros
  • Lower risk of AI shaping the design
  • Easier to review individual outputs
Cons
  • Leaves most production speed unused
  • Does not exploit AI for research, synthesis, or structured artifact generation

The useful operating model was not AI autonomy. It was governed acceleration. AI made speed possible, but the exercise still needed human authority to define what good looked like, reject unsuitable outputs, and make design changes from execution evidence.

Separating generation from authority

This decision is the core of the CADE showcase: the AI skill is not merely prompting. It is deciding where AI can create leverage and where human judgment must remain authoritative.