An AI system that supports critical processes has deviated from expected performance and is producing biased outcomes. Which of the following is the BEST course of action?

The correct answer is B. Perform a root cause analysis to identify mitigation steps. Performing a root cause analysis (RCA) is the best first step because it identifies exactly why the bias is occurring before any corrective action is taken. Without understanding the root cause, any remediation may be ineffective or incomplete. Retraining the model with new data

AI Security Operations and Monitoring

Question

Options

ARetrain the model with a new and expanded dataset
BPerform a root cause analysis to identify mitigation steps
CConduct audits of the data and the model
DActivate the model kill switch

How the community answered

(27 responses)

A
4% (1)
B
81% (22)
C
4% (1)
D
11% (3)

Explanation

Performing a root cause analysis (RCA) is the best first step because it identifies exactly why the bias is occurring before any corrective action is taken. Without understanding the root cause, any remediation may be ineffective or incomplete. Retraining the model with new data (A) could help, but if the root cause is not identified first-such as a data pipeline error, a flawed feature, or a model architecture issue-retraining may not fix the problem or could introduce new issues. Conducting audits (C) is valuable and may be part of the RCA process, but auditing alone does not produce an action plan. Activating the kill switch (D) is a last resort for immediate safety threats; it is overly disruptive for a situation that warrants investigation and measured remediation first.

Topics

#AI bias#Root cause analysis#Incident management#AI system performance

Community Discussion

No community discussion yet for this question.

Full AAISM Practice