Formative assessment

Formative assessment gathers feedback during learning so teachers and learners can adjust what happens next. It often has low or no point value. Summative assessment evaluates learning at the end of a unit against a standard, although that feedback may later inform learning too. Carnegie Mellon's overview explains the distinction by purpose and use.

For an AI-assisted workflow, keep two questions separate: is the draft activity sound, and how will the educator use responses? Reviewing the first question does not authorize a model to answer the second for an individual learner.

Consider a fictional map lesson teaching two facts: the handout's dotted line marks a walking path, and its north arrow indicates north. A short exit ticket asks about each fact. If the generated key says the path uses a solid line, fix the key before giving the ticket to learners. That error belongs to the draft and is not evidence of a learner misconception.

The exit-ticket workflow maps each prompt and expected answer to an approved objective and taught passage. A drafting surface such as MagicSchool can be part of that preparation, while the educator decides whether wording and answers fit the lesson.

An LLM eval dataset measures an application's behavior against test cases. A formative activity serves teaching and learning. The Exit Ticket Reviewer checks supplied draft materials for the latter workflow; it does not score learners. After classroom use, the educator may choose a clarification or follow-up activity. Keep that judgment with the educator, and keep questions about untaught material out of the reviewed ticket until the source gap is resolved.