In an AI risk management context, what does 'concept drift' refer to?
Select an answer to reveal the explanation.
Short Explanation and Infographic
Here's the deal — b is correct because concept drift refers specifically to the change in the relationship between input features and the outcome variable (P(Y|X) changes over time), meaning the patterns the model learned no longer accurately describe current reality. For example, a model predicting loan default may have learned patterns from a pre-recession economy that no longer hold post-recession.
Full explanation below image
Full Explanation
B is correct because concept drift refers specifically to the change in the relationship between input features and the outcome variable (P(Y|X) changes over time), meaning the patterns the model learned no longer accurately describe current reality. For example, a model predicting loan default may have learned patterns from a pre-recession economy that no longer hold post-recession. Changes in input distribution without relationship change (A) is data drift. Software bugs (C) are implementation issues. Expert judgment divergence (D) is not a standard ML definition.