Human agency and oversight keeps people meaningfully involved in an AI-enabled system. Treat it as a delivery and governance requirement: preserve insight, interaction, and the ability to disable predictive analytics. section 6.6.2
Convert the principle into sprint commitments:
- Address oversight in sprint 2 by selecting modelling techniques that support user insight and interaction. section 6.6.2
- Add controls that disable predictive analytics when necessary. section 6.6.2
- Offer trend-only or steady-state forecasts as safer operating alternatives. section 6.6.2
- Record the decisions and communicate them through an ethics report. section 6.6.2
The Bike Shop predictive analytics application
An ethics assessment identifies human agency and oversight as critical alongside data governance. The team responds by planning interactive modelling tools, disablement controls, and less aggressive forecast modes before evaluation proceeds. section 6.6.2
Figure 6.9 Ethics assessment matrix for The Bike Shop, based on an IEEE use case matrix
Use the ethics assessment matrix to locate human agency and oversight among the application's critical concerns. section 6.6.2
