AI Glossary Term

What does “Model Drift” mean?

Definition

Performance degradation over time as real-world conditions diverge from training assumptions.

Browse the full AI Glossary Explore all free AI guides

How Model Drift works

Example

A fraud model works well at launch. Months later, fraudsters change tactics, and the model's catch rate falls. Monitoring shows the decline, and the team retrains on newer examples.

Why it matters

A model that is not monitored can quietly get worse. Regular checks protect against stale or failing predictions.

Common questions

What causes model drift?

Changes in inputs, in the relationship between inputs and outcomes, or in how people use the system.

How is it detected?

By tracking performance on labeled samples and comparing input statistics with the training data.

Is drift always bad?

It signals the model may no longer fit the world. Small drift may be tolerable, but large drift needs action.

Data DriftA shift in real-world input data over time that can degrade model performance.Machine Learning (ML)Methods that allow systems to learn patterns from data and improve over time.Training LossThe model error value computed during training and optimized downward over time.Autonomous SystemA system that can make decisions and act with limited or no direct human control in real time.Continual LearningTraining approaches that let a model keep learning from new data without forgetting prior knowledge.Knowledge CutoffThe latest point in time reflected in a model's training data.

Learn more in our free guides

AI Models Explained Model Collapse Model Lifecycle Model Context Protocol

Test yourself

AI Models Explained Quiz Model Collapse Quiz Model Lifecycle Quiz