Why are low-effort explanations excluded from AI insights?

The short answer

Because a summary built from twenty explanations, six of which say “idk”, is a worse summary than one built from the fourteen that say something.

Stock non-answers dilute the themes and distort the counts. Leaving them out of the summary makes what you get back more useful.

Nothing is deleted

Every response is still there and you can still read all of them individually. The exclusion applies only to the set the AI summary is built from.

The insights card tells you how many were set aside, so a shrinking total never looks like responses going missing.

What counts as low effort

Not a judgement about quality. It catches:

• Stock non-answers, such as “idk” or “no idea”
• Random keystrokes
• Punctuation, symbols or emoji with no words in them at all, such as ”?”, ”…” or a lone shrug

Words in any language count as words. An explanation written in Irish, Japanese or Arabic is not low effort. That was a deliberate decision, and it matters for any school with students writing in a language other than English.

A short answer is not automatically low effort either. “Because it doubles” is brief, but it says something, and it will be included.

On multiple choice questions

Where a student picked an option and wrote a throwaway explanation, the explanation is left out of the summary but the answer itself still counts. You do not lose the response.

Can I overrule it?

Yes. You can change how any explanation has been treated, and your decision stands.

If the summary looks out of date

Summaries are tied to the set of responses they were built from. If that set changes, for instance because more students submit, you will be offered a fresh summary rather than shown the old one as though it were current.