Subject
Computer Science
Topic
Artificial Intelligence
A system scores 99% accuracy because the rare harmful class is almost absent. What should be examined?
A system scores 99% accuracy because the rare harmful class is almost absent. What should be examined?
a)
Only the total sample countb)
Only model file sizec)
Only training durationd)
Class-specific metrics and the confusion matrixExplanation
Accuracy can hide failure on a minority class, while precision, recall, and class counts expose the relevant errors.Related questions
A loan model rejects qualified applicants from one demographic more often than others. What should be investigated first?A hospital uses AI recommendations but requires a doctor to approve every treatment. Which governance pattern is this?A payroll script follows fixed rules and never learns or reasons about alternatives. What is the safest description?A vision model performs well in training but fails after camera lighting changes. Which general AI risk is illustrated?A company collects unnecessary personal fields for an AI feature. Which privacy principle should guide redesign?A bank must explain why an automated decision was denied. Which property is most directly required?A symbolic planner and a learned vision model are combined in one robot. What is this approach commonly called?An AI capability works in a laboratory but lacks monitoring after deployment. What important lifecycle activity is missing?