101
Start free
Building AI systems

Evaluation

Level 2

Systematically testing whether an AI system's output is actually good.

Term 4 of 6 in Building AI systems

In plain language

Evaluation, often shortened to evals, means running a fixed set of test cases and scoring the results. Without it, "the new model feels better" is the only evidence anyone has.

Think of it like this

A tasting panel with a scorecard instead of one chef's opinion.

Why it matters

Evals are what separate teams that improve their AI feature from teams that just change it.

Next terms in Building AI systems