AI Use rubric builder
A rubric is a shared decision tool that helps people make similar choices in similar situations. When AI is involved, a rubric also makes responsibility visible, so decisions don't live inside individual workflows where others can't find or follow them.
In this unit, you learn the building blocks of an AI Use rubric and practice drafting rubric language that works for both classroom and staff contexts. You leave with a starter set of rubric rows you can build on in the practice experience. This is one of the most transferable tools you create in this module.
Key ideas and models
Let's explore some key ideas and models.
What an AI Use rubric does
An AI Use rubric translates values into observable expectations. It makes it easier to decide what fits, what requires oversight, and what shouldn't be delegated. It also supports transparency because teams can point to shared criteria rather than personal preference. A rubric is most useful when it reduces guesswork. It should help someone answer: what is the task, what is the risk, what must stay human, and what checkpoints must happen before anyone relies on the output?
The rubric row template
A practical rubric uses rows and columns. Each row is a task type. Each column is a decision element that keeps responsibility visible. A strong row can be reused across tools because it focuses on the decision, not the specific technology.
Use this row template:
- Task type
- Category: appropriate, appropriate with oversight, or not appropriate
- What must stay human
- Privacy default
- Required checkpoint before use or sharing
- Transparency line
- Documentation to keep
Classroom and staff contexts use the same logic
Classroom AI decisions often center learning, student agency, and integrity. Staff AI decisions often center trust, accuracy, privacy, and accountability in communication and records. The same rubric structure works for both when the row language names who relies on the output and what must be checked. A practical rule is to increase oversight when the output shapes opportunities for learners or becomes part of an official message or record. Reduce data sharing by default, especially when a draft could include student identifiers or sensitive details.
Writing rubric language that people can follow
Rubric language works best when it is specific and observable. Avoid words that depend on personal interpretation, such as "safe" or "appropriate," without naming what must be done. Use verbs that describe action: verify, remove identifiers, revise for tone, and document decisions. A helpful test is to ask whether a new staff member could follow this without guessing. If not, add one checkpoint or one example to make the expectation clear.