Drift monitoring
The ongoing tracking of whether an AI system's real-world performance is holding steady or degrading over time, distinct from a one-time accuracy or fairness test at signing. Belongs in the operational and monitoring evidence category because AI systems can change or drift after deployment even without either party misrepresenting anything at the outset.
Defined in 2 GAGE programs, which carry 2 distinct definitions of it. The wording above is taught in Certified AI Transformation Professional (CATP).
How each discipline defines it
The same term does different work depending on who is using it. These are the definitions as each program teaches them, unedited.
The ongoing tracking of whether an AI system's real-world performance is holding steady or degrading over time, distinct from a one-time accuracy or fairness test at signing. Belongs in the operational and monitoring evidence category because AI systems can change or drift after deployment even without either party misrepresenting anything at the outset.
Periodic sampling of AI outputs and the team's verification discipline to catch the quiet decay of both quality and checking habits over time, so review can be re-tightened when a trigger signal appears (such as a two-consecutive-week rise in error rate or a verification-step completion drop from the established normal).
Where it is taught
The exact lessons this term appears in. The first module of every program is free with a free account.
- Managing AI-Augmented Teams: The Middle Manager's Role · AI in the Workplace and Team Leadership, Certified AI Practitioner: Workplace Foundations
- Negotiating Evidence Rights You Will Actually Use · Technology, Platforms and Vendors, Certified AI Transformation Professional (CATP)
Terms it appears with
Not an alphabetical neighbourhood: these are the terms taught in the same lessons, ranked by how often they appear together.