AI glossary
Model drift
Model drift is a change in how well an AI feature works over time, because the inputs, the world or the model behind an API has changed.
Data drift is when users start asking different things; model drift in the wider sense includes a provider updating the model behind the same name.
You notice it by watching: run evals on a schedule, track quality signals from users and compare against a baseline.
Pin model versions where you can and change them on purpose, so a silent update does not change your product overnight.
Learn it hands-on
These free lessons teach it with commands to run in a lab and checked tasks. Read them whole with no account.