Degradation of an AI model's performance over time as real-world conditions change from what it was trained on. Like skills becoming outdated as the world changes.
A fraud detection model experiences drift as scammers develop new tactics that weren't in the training data, requiring periodic retraining.
All providers support detecting changes in production data and model behavior over time. AWS SageMaker Model Monitor, Azure ML data drift monitoring, and Vertex AI Model Monitoring provide managed drift/monitoring features. OCI supports model deployment and operational monitoring, but drift detection is typically implemented using custom logging/metrics and alerts (often combined with OCI Monitoring/Logging) rather than a single dedicated drift product.