Evaluation & Quality
Robustness
How consistently a system performs when inputs, wording, or conditions change.
Example
Small wording changes should not cause a reliable classifier to reverse its decision.
Why it matters
It helps teams judge whether a tool is dependable beyond a narrow demonstration.
You'll hear it
“It should still work when the request is phrased a little differently.”
Can you control it?
Not directly
No, not directly. Robustness describes a learned property, observed result, or broader concept rather than a normal user setting. You may influence it indirectly by changing the tool, data, prompt, or workflow.
Common questions
- Does robustness mean the system never fails?
- No. It concerns performance across specified changes, not every imaginable situation.
- Can a system be robust to noise but weak on new topics?
- Yes. Different kinds of change test different abilities.
- Can a small change have a large effect?
- Yes. Some systems are sensitive to wording, formatting or unfamiliar details.