An AI auditor reviewing a model's post-deployment monitoring program finds that alerts are configured only for API error rates and latency. What significant monitoring gap does this reveal?
Select an answer to reveal the explanation.
Short Explanation and Infographic
Here's the deal — a is correct because monitoring only API errors and latency captures infrastructure health but misses model quality — specifically whether the model's predictions remain accurate and whether input distributions are shifting. Comprehensive AI monitoring requires model performance metrics and data drift detection.
Full explanation below image
Full Explanation
A is correct because monitoring only API errors and latency captures infrastructure health but misses model quality — specifically whether the model's predictions remain accurate and whether input distributions are shifting. Comprehensive AI monitoring requires model performance metrics and data drift detection. B is a security monitoring gap. C and D are not standard model monitoring requirements.