Staff must learn when to trust an agent's output. What belongs in adoption planning?
Why C is the answer
Training on what it can and cannot do
Adoption depends on clear guidance and sensible human oversight.
In detail
- Staff need to know what it can and cannot do.
- Teach people when to escalate.
Use case
An illustrative scenario, not a real incident.
Staff are shown where an agent is strong and where it makes mistakes, so they check what matters.
See it in different ways
stateDiagram-v2 [*] --> Pilot Pilot --> Measure Measure --> Improve: low ratings on a topic Improve --> Measure Measure --> Expand: targets met Measure --> Rollback: serious problem
mindmap
root(("Adoption training"))
n1["Can do"]
n2["Strengths"]
n3["Cannot do"]
n4["Limits"]
n5["When to check"]Key point: Train people on the limits.
Why the other options are wrong
- A. A policy that its output is always right
- That is false, and it is dangerous.
- B. A longer list of the agent's features
- Features do not teach judgment.
- D. A rule against asking it any questions
- A ban means no adoption.