Skip to content

Evals, evaluation

Adrià Castany 2 min

What is an evaluation?

An evaluation is a set of test cases used to check, repeatably, whether an AI system answers as it should. Each case carries an input and a success criterion, and the whole set is re-run every time something changes. The prompt, the model or the documentation.

Also called

evals, evaluations

Topic
AI in support
On this page
  1. Testing by hand is not evaluating
  2. Why it matters
  3. Where the cases come from
  4. What gets checked
  5. When they are run
  6. How Intake handles it
  7. Frequently asked questions

Testing by hand is not evaluating

Typing four questions into the chat after a change and seeing whether it answers well is a check, not an evaluation.

What makes it an evaluation is that it is always the same set, so results can be compared. Without that you cannot tell whether the change improved anything or just moved the problem.

Why it matters

Because in an AI system any change can make worse things that already worked, and you cannot see that by looking.

Adjusting the prompt to fix one answer and breaking ten others is normal without evaluations, and the worst part is that nobody finds out until a customer says so.

Where the cases come from

From your own conversations.

The fifty most frequent questions, plus the cases that went wrong, plus the limits you want to protect, meaning what it must not answer. Inventing generic cases gives you an evaluation that always passes and protects nothing.

What gets checked

Three things in each case.

  • That the answer contains the right fact.
  • That it does not contain what it must not, such as an invented figure or a promise.
  • That it hands the conversation to a person when it should. It is the most forgotten and the most expensive when it fails.

When they are run

Before every change that reaches customers.

It is the same discipline as tests before a deploy, applied to something non-deterministic, which is why it needs it more.

How Intake handles it

Frequently asked questions

How many cases does an evaluation need?

Start with thirty or forty taken from your own conversations. That is enough to catch a regression and few enough that somebody will maintain them.

Can answers be evaluated automatically?

Partly. Checking that a specific fact appears, or that something forbidden does not, automates well. Judging whether an answer is well written still needs judgment, even when another model is used to score it.

What is the difference between an evaluation and a software test?

A software test always expects the same result. An evaluation accepts answers worded in different ways and checks whether they meet a criterion. That is why part of the marking is done by a person or by another model.

Glossary

Keep reading

Glossary

Related terms

See the full glossary

Add it without dropping what you use

15 minutes, no commitment