Formative Assessment Prompts: Quick Checks That Actually Inform

A formative question like 'what is photosynthesis?' measures recall, not understanding. The most useful formative checks expose what students believe that is wrong — misconceptions the next lesson must address. AI can generate hinge questions and misconception traps if you tell it what common errors look like.

Formative probe types

Closing the loop

Formative data is useless if it only travels from student to teacher. Show students what the class's answers revealed — anonymously — at the start of next class. That meta-transparency is itself a learning moment.

The big idea: the best formative check isn't the hardest question — it's the question that exposes the most common wrong belief.

Formative assessment cadence is constrained by authoring overhead; AI removes the constraint without replacing the educator's judgment about what to assess.

What AI does well here

• Generate questions aligned to specific learning objectives at specified Bloom's levels
• Produce variations of each question to create equivalent forms
• Draft answer-explanation feedback for each question
• Generate misconception-targeted distractors that surface common errors

What AI cannot do

• Substitute for the educator's selection of which objectives to assess when
• Replace formative-assessment data analysis (the educator interprets and acts)
• Generate authentic performance assessment items (those need real-world tasks)

Formative assessment depth is limited by teacher grading bandwidth; AI handles the grading so teachers focus on response.

What AI does well here

• Generate quizzes aligned to today's lesson, not generic content
• Auto-grade and surface patterns (where the class is stuck)
• Adjust next-day instruction based on quiz signals
• Maintain student-friendly framing (formative, not punitive)

What AI cannot do

• Substitute for the relational reading of student understanding
• Replace teacher judgment about what to do with the data
• Eliminate the value of the in-the-moment classroom check

Formative checks default to comprehension questions; AI drafts richer items aligned to objectives.

What AI does well here

• Draft items at varied DOK levels
• Align items to learning objectives you supply
• Suggest follow-up questions per item

What AI cannot do

• Predict actual student difficulty
• Substitute for teacher knowledge of your class

Designing Richer Formative Checks With AI Assistance

Most formative checks default to recall questions because they're fast to write. AI makes DOK Level 3 and 4 items — application, analysis, argument — just as fast to draft as recall items. The key is telling AI exactly what you want. Try: 'Given this learning objective: Students will analyze the causes of World War I. Draft 6 formative items at mixed DOK levels — 2 at DOK 1 recall, 2 at DOK 2 skill/concept, 2 at DOK 3 strategic thinking. Include answer keys and one follow-up probing question per item.' You get a tiered item bank ready for exit tickets, quick polls, or a structured class discussion. The follow-up questions are particularly valuable — they're the next question a skilled teacher would ask a student who just answered, and they are the hardest items to write under time pressure. AI drafts them in seconds.

• Specify DOK level distribution when requesting formative items (e.g., 2 DOK 1, 2 DOK 2, 2 DOK 3)
• Always include the specific learning objective in your AI prompt
• Request follow-up probing questions for each item to support classroom discussion
• Pilot new items with one class before deploying widely — revise based on actual responses
• Build a shared team item bank organized by standard and DOK level

Common wrong answers reveal common misconceptions; AI clusters them so reteaching targets the right one.

What AI does well here

• Cluster wrong answers by likely misconception
• Suggest the next-day reteach that targets the most common gap
• Surface students whose error suggests they need a different intervention

What AI cannot do

• Replace the teacher's diagnostic conversation with a confused student
• Know which kids are guessing vs. genuinely confused
• Tell you what the right reteach actually is

Building enough quality formative items per standard is grinding work. AI can produce serviceable drafts at scale — provided you review for bias, clarity, and alignment.

What AI does well here

• Generate 10 items per standard at varying DOK levels
• Vary item formats (MC, short response, performance task)
• Draft answer keys with common misconceptions
• Tag items by standard and difficulty

What AI cannot do

• Verify cultural responsiveness of items
• Replace a teacher's read on what students are actually missing
• Validate items against your specific student population
• Substitute for vetted assessment products in high-stakes contexts

AI can draft formative-assessment question banks where wrong answers map to specific known misconceptions, turning the item into diagnostic data.

What AI does well here

• Generate items where each distractor maps to a documented misconception.
• Draft exit-ticket variants at 3 difficulty levels for the same learning target.

What AI cannot do

• Replace the teacher's read of which misconception this specific class actually holds.
• Substitute for the conversation that addresses the misconception once revealed.

AI can generate quick formative checks for any unit, but the real responsive teaching happens in how you read the answers and adjust today.

What AI does well here

• Generate 3 quick checks per learning target
• Suggest exit tickets in multiple formats
• Build a 5-minute response plan for common errors
• Draft a Friday-review review prompt for next week

What AI cannot do

• Tell you which student needs which intervention
• Replace your read of the room
• Predict misconceptions you have not seen yet