Leadership checklist
Responsible AI adoption: a leadership readiness checklist
Before approving a tool or pilot, leadership should be able to answer these twelve questions in the institution’s own operating language.
The readiness test
A “yes” to AI is not a single decision. It is a set of decisions about purpose, information, accountability, review, capability, and value. Unresolved answers do not necessarily mean the institution should stop. They identify the work that must be completed before responsible implementation.
- 01
Purpose
Can we name the operating problem or service opportunity without referring to a specific AI tool?
- 02
Value
What observable outcome should improve, and what baseline will we use?
- 03
Ownership
Which executive owns the result, and who owns the workflow day to day?
- 04
Information
What information enters the system, and what information must never be entered?
- 05
Decision rights
Which decisions remain human, and who has authority to approve or override output?
- 06
Accuracy
How will the team identify, review, and correct unreliable output?
- 07
Fairness
Could the workflow create different consequences for clients, members, staff, or communities?
- 08
Transparency
Who should know that AI is being used, and what should they be told?
- 09
Vendor risk
What does the supplier do with institutional information, and how can the arrangement be exited?
- 10
Capability
Do the people responsible for the work understand the process, limits, and escalation route?
- 11
Measurement
How will leadership know whether the workflow is adopted, safe, and producing value?
- 12
Stop condition
What evidence would cause the institution to pause, redesign, or stop the use?