Leading AI at Work: Governance, Rollout and Practice / Unit 3 LAI 060 worked example, 60 to 90 seconds
The Gold Thing Safety Checklist
The idea
Before changing a Gold thing or adding an AI feature around one, the book asks six questions.
In practice
Step one: is there a Gold Pack, and is it current? Step two: is there a traceability matrix, and has it been updated for this change? Step three: have you identified which fields the AI reads, and which fields or states it writes or influences? Step four: is the thing's purpose written down in its Pack? Step five: are the human review points, the rollback options, and the logging and audit for AI-affected changes clear? Step six: have you generated test scenarios from the cards that include the AI's behavior?
Where people slip
The questions are cheap to ask and expensive to skip. The fifth one carries the most weight, because review, rollback and audit are what let a team recover when the AI behaves in a way nobody planned.
Your place on the trail
Every lesson rests on earlier ones and opens later ones. Tap any to jump.
Something wrong in this lesson? Flag it · Sources for Leading AI at Work: Governance, Rollout and Practice