Role library · aec-roles-v1 · technology
AI Exposure Check
Quality Assurance Engineer
Illustrative task-level Exposure from the offline sample model — not a market survey.
Exposure 56% average exposure · CHF 107136 illustrative time value · 5 tasks
Tasks
Exposure and reasoning
| Task | Exposure % | Reasoning |
|---|---|---|
| Write boilerplate CRUD endpoints | 73% | Moderate exposure: “Write boilerplate CRUD endpoints” mixes templateable steps with contextual judgment; AI can draft or triage, but an accountable human should decide. |
| Generate unit-test scaffolds | 74% | Moderate exposure: “Generate unit-test scaffolds” mixes templateable steps with contextual judgment; AI can draft or triage, but an accountable human should decide. |
| Summarise incident postmortems | 63% | Moderate exposure: “Summarise incident postmortems” mixes templateable steps with contextual judgment; AI can draft or triage, but an accountable human should decide. |
| Review pull-request diffs | 50% | Lower-moderate exposure: “Review pull-request diffs” requires coordination or domain nuance; AI may summarise inputs but should not own the outcome. |
| Pair on novel architecture | 19% | Low exposure: “Pair on novel architecture” is relationship-, safety-, or judgment-critical; keep a human owner and use AI only as a narrow aid. |
Sample actions
- Pilot generative drafting on the two highest-exposure tasks with human review.
- Bundle similar low-judgment tasks into one assisted workflow queue.
- Set an audit sample rate for AI outputs before customer or regulator exposure.
What these figures are not
- Illustrative role library from the offline AEC sample model (aec-ai-v1), not a surveyed occupation census.
- Task packs and exposure scores are deterministic samples for self-serve triage — Faethm-style ontology breadth, not Pearson proprietary data.
- CHF time-value estimates use illustrative Swiss mid-market rates and annual hours.