Which of the following is the MOST critical key risk indicator (KRI) for an AI system?

The correct answer is D. The rate of drift in the model. Model drift - the gradual degradation of a model's performance as real-world data patterns diverge from training data - is the most critical KRI for an AI system because it is a leading indicator of systemic failure. While accuracy rate (A) measures current performance, it is a l

AI Security Risk Management

Question

Options

AThe accuracy rate of the model
BThe amount of data in the model
CThe response time of the model
DThe rate of drift in the model

How the community answered

(35 responses)

A
6% (2)
B
3% (1)
C
9% (3)
D
83% (29)

Explanation

Model drift - the gradual degradation of a model's performance as real-world data patterns diverge from training data - is the most critical KRI for an AI system because it is a leading indicator of systemic failure. While accuracy rate (A) measures current performance, it is a lagging metric; it tells you something already went wrong. The amount of data (B) is a resource metric, not a risk indicator. Response time (C) reflects performance but not risk to output quality. Drift (D), by contrast, signals that the model's predictions are becoming less reliable over time, which can compound into significant business, regulatory, or safety consequences if not caught early.

Topics

#Key Risk Indicators (KRIs)#AI System Risks#Model Drift#AI Monitoring

Community Discussion

No community discussion yet for this question.

Full AAISM Practice