Glossary
Human baseline
What is a human baseline?
A human baseline measures how well people do the same tasks an AI system is tested on, with the same cases and the same definition of correct, so the two can be compared fairly.
Why it matters
Without a careful baseline, you can’t tell whether an agent beats your team or only beats a guess. People should be timed and scored as carefully as the AI.
Example
Before testing an agent on 40 past refund requests, a team times two experienced specialists on the same 40 and scores everyone with one rubric.