AI Integration
:::info Source
Sourced from services/enrollment-service/AI_INTEGRATION.md in the documentation repo.
:::
1. Scope
No direct AI in enrollment-service. Two passive touchpoints:
- AI-suggested enrollments (via assignment-service; enrollment just receives events).
- AI-driven retention nudges (analytics-service) reference enrollment data.
2. Data Exposed to AI Consumers
- Aggregate enrollment counts per course / cohort.
- Enrollment state transitions for retention models.
3. Safety
- Actor IDs hashed before analytics firehose.
noTrainon downstream providers.
4. Provenance
N/A (no AI generation here).
5. Bias
- Enrollment data is neutral input; bias concerns live with consumers (analytics AI).
6. Right to Explanation
- Learners can see why they are enrolled (source.kind) — "purchase order ord_…", "assignment A-123", etc.