category: Assessment
---
Formative assessment is the most powerful thing you can do in a classroom. The idea is simple: constantly check what students actually understand, use that to shape instruction the next day. But the logistics kill you. You have 28 kids. There's no way to read every draft, every response, every piece of thinking every day. So you sample, and you hope you catch the patterns.
AI changes the equation here. It can't replace your judgment about what matters, but it can be an exhausted teacher's reading buddy—someone who reads everything, flags patterns, and gives you the summary you need to decide what to teach tomorrow.
The pattern-spotting move
Here's what works: you collect your students' exit tickets, drafts, responses from class discussion notes—whatever evidence you have of their thinking. You paste it into an AI with a specific instruction:
"I have exit tickets from 27 students on [concept]. Read all of them. For each major misconception or gap you see, count how many students showed it and give me 2-3 examples. Order them by frequency—what's the most common misunderstanding? Then give me one sentence: what should I reteach tomorrow?"
You get back something like:
- 8 students think photosynthesis is just about plants getting energy from soil (you get their exact responses, not a summary)
- 5 students didn't attempt the open-ended question at all
- 3 students over-generalized ("all green plants, all the time")
And your reteach decision makes itself. The AI didn't teach anything. It read faster than you could and gave you signal.
What you still do
- You read the examples it pulls. The AI might mischaracterize a response, so you check the actual student work.
- You make the instructional decision. "OK, misconception #1 is the biggest problem. Tomorrow I'm starting with that." That's you. That's your professional judgment.
- You design the reteach. The AI can suggest ideas, but you know your kids and what moves they need.
The drafting version
Same move for written work. A student drafts an essay. You want feedback on content, not just surface errors. But giving genuine content feedback on 28 drafts takes hours.
Instead:
- Student submits draft
- You paste it into an AI: "This is a draft on [topic]. Give me 3 sentences of feedback: what's the strongest thinking here? What's the biggest gap? What question would push this writer's thinking forward?"
- You read that feedback. Does it match what you see? If yes, share it with the student. If it's off, you write your own.
Now you've done genuine formative assessment without the time kill. The AI is doing the reading; you're doing the judgment.
The limits (and they're real)
- AI is bad at understanding writing that's incomplete or using unusual structures. If a student's thinking is non-linear or fragmented, the AI might read it as confused when it's actually sophisticated.
- It misses genre and context. An AI doesn't know if this is meant to be formal or conversational, if the student is taking a risk with voice, if they're being intentionally informal.
- It can't hear culture. Phrasing that's perfectly valid in a student's home context might read as "unclear" to an AI trained on standardized English.
You HAVE to read it. The AI is saving you time on pattern-spotting, not replacing your eye.
The practical setup
Here's what this looks like in a real classroom:
Monday: Exit tickets on the day's learning. You paste all 27 into an AI at 3 PM. By 3:15, you have a summary of misconceptions.
Tuesday morning: You know what to reteach. You design it. It's way more targeted than if you'd skimmed 27 papers and hoped you caught the patterns.
Mid-unit drafts: Student submits. AI gives 3-sentence feedback. You read both the draft and the feedback, decide if it's useful, and share or write your own.
End-of-unit assessments: You're looking for trends, not individual errors. AI reads 27 essays and tells you: "12 students still confusing X and Y. 8 students nailed the analysis. 7 didn't attempt the synthesis question." You see what to reteach or dive into with individuals.
One more thing: transparency
Tell students this is happening. "I use AI to help me read faster so I can give you better feedback." They don't need to think you're spending hours on their work if you're using tools smartly. That's not cheating; that's being efficient with the one resource you don't have: time.
Real talk
Formative assessment without the AI is a job you can't actually do well at scale. You read what you can, you miss patterns, kids don't learn the thing they didn't understand because you didn't notice they didn't understand it. With AI reading alongside you—and you doing the judgment—you can actually do this job the way it's supposed to be done.
That's not replacing teaching. That's finally having time to teach.
