Leading AI at Work: Governance, Rollout and Practice / Unit 6 LAI 144 worked example, 60 to 90 seconds
Planning to Switch It Off
The idea
NIST's Manage two point four asks for mechanisms, with assigned responsibilities, to supersede, disengage or deactivate AI systems whose outcomes are inconsistent with intended use.
In practice
Step one: for each AI feature, write down the signal that would show its outcomes are inconsistent with its stated purpose. Step two: name the person with the authority to disengage it, and a deputy. Step three: describe the fallback, the way work continues without the feature, and test it once before launch. Step four: record in each affected Gold Pack what happens to records the feature has already changed. Step five: add the plan to the feature's matrix row, so that anyone walking a change can find it.
Where people slip
The book's checklist asks the same thing in plainer words: how will you roll back or adjust? A fallback that has never been tried is a hope, not a mechanism.
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