Data Drift

A change in the statistical distribution of input data over time, which can affect model performance even if the relationship between features and target remains stable.

When the type of data coming in changes from what the AI was trained on. If a retail AI was trained on pre-pandemic shopping data, post-pandemic shopping patterns might confuse it.

Data drift is an operational risk that can undermine AI performance without warning. Your AI monitoring and governance processes should include automated drift detection and clear escalation procedures when data distributions shift significantly.