Role-on-the-Wall
Draw the outline of one role most affected by AI on a large sheet. Inside the outline goes the human core — what this person knows, decides and is accountable for. Outside goes the AI boundary: what AI does, what it should support, what it must never touch. The line between them is your standard.
Use when a team needs to write an AI standard for a role but keeps producing vague policy language.
Solves: Blanket AI policies say "be careful", which operators read as "do not use it for anything important".
- 1
Step 1: Pick the one role most affected by AI. Draw its outline, life-size if you can, on paper on the wall (5 min).
- 2
Step 2: Inside the outline, write what this role knows, decides, and is accountable for (10 min).
- 3
Step 3: Outside the outline, write three columns: AI does, AI should support, AI must never (15 min).
- 4
Step 4: Trace the line between inside and outside. Write one sentence describing how the human and AI collaborate. That sentence is the seed of the tiered standard (10 min).
- Insist the role is a real named role in the organisation, not an abstraction.
- The 'AI must never' column is the most valuable and the one groups rush. Slow it down.
- If people write tasks inside the outline, redirect: inside is judgement and accountability, not activity.