OPTIONAL PRACTICAL EXTENSION · HUMAN ASSESSED
Create a report you can stand behind.
This task complements the SCORM knowledge check. A facilitator scores the evidence with the practical rubric from the AI at Work outline. This score is separate from the automated course score.
The brief
You are preparing a one-page service update for the operations manager. Explain what changed, identify limits of the evidence and recommend three possible next steps. Use the fictional data below. Do not upload real workplace information.
| Metric | April | May |
|---|---|---|
| Requests received | 400 | 500 |
| Average first-response time | 10 hours | 9 hours |
| Requests resolved within the service target | 320 of 400 | 425 of 500 |
| Customer satisfaction | Not collected | Not collected |
The data does not include staffing, case complexity or reasons for the change. Do not infer causes or customer satisfaction from it.
Produce your evidence
- Choose a tool your organization permits for these fictional inputs. If none is available, draft your prompt and report manually and mark tool use as not demonstrated.
- Record your task, audience, success criteria, tool choice and limitations.
- Save your first prompt and output. Revise the prompt after reviewing the result, then explain what improved.
- Check the material numbers and claims against the table. Keep a short verification log showing calculations, corrections and unresolved gaps.
- Edit the final one-page report. Include limitations, three possible next steps and the human review checkpoint.
- Map the workflow before and after AI support. Propose a small trial with a measure such as time saved and number of corrections. Do not claim implemented automation unless you demonstrate it.
Practical rubric · 18 points
For each criterion: 3 = independent and complete; 2 = main requirements met with minor omissions; 1 = partial evidence or substantial guidance; 0 = no usable evidence.
| Criterion | Evidence for full achievement | Score |
|---|---|---|
| Task and tool selection | Useful task, audience, success criteria, justified tool choice and limitations. | __/3 |
| Prompting and refinement | Context, constraints and output requirements; revised prompt and reason for the revision. | __/3 |
| Accuracy and validation | Claims and calculations checked against the inputs; corrections and uncertainty recorded. | __/3 |
| Responsible use | Permitted tool and fictional data; relevant bias/copyright considerations and accountable human review. | __/3 |
| Usable final output | Report meets the brief and audience needs, with documented quality checks. | __/3 |
| Workflow improvement and transfer | Feasible before/after workflow, human checkpoint and measurable trial. | __/3 |
Suggested pass: 14/18, including at least 2/3 in both accuracy and validation and responsible use. A material privacy breach or unchecked consequential error requires correction and reassessment. This is a proposed course criterion, not an externally validated certification.
Facilitator verification key — open after attempting
Requests increased by 25%. Average response time fell by 10%. Within-target resolution rose from 80% to 85%, an increase of 5 percentage points. Causes and customer satisfaction are unknown. Recommendations should be framed as proposals requiring investigation.