Metrics for GenAI results · Question 50 of 60
When compared with an expert suite, some AI test cases count as wrong although they check sensible cases that the experts missed. What does this show?
1 answer is correct. Hard Created with AI
Show answer
Answer
- A An automatic comparison is only as good as the reference; a manual review remains important
Explanation
An incomplete reference also penalizes useful deviations. People recognize such cases; at the same time, the reference can be extended with the new tests.