When AI Writes It For Them: Assessing the Process, Not the Product

The assignments that come in these days are uniformly polished. The sentences are tidy, the structure is logical, there are no typos. And yet call the student over and ask "why did you write it this way here?" and the answer stalls. Now that generative AI turns out a plausible finished product in seconds, it has become hard to believe that a highly polished submission equals a student's understanding. But throwing yourself into hunting down and punishing AI use is a losing game. The direction should be the opposite: pull the process that leads to the product, rather than the single finished page, into the scope of assessment.

Why assessment that looks only at the product wavers

Grading only the finished output now reveals a few gaps.

• You can't see who made it: a product shows only its final state; there's no way to know whether the student arrived at it themselves.
• Polish and understanding come apart: smooth writing doesn't necessarily mean deep understanding. If anything, the more it's polished, the more the student's traces are erased.
• The places to give feedback disappear: when the wrestling, the revising, and the questions leave no trace, the footholds for growth a teacher could point to disappear with them.

The point is simple. Move the center of gravity of assessment from "what did you turn in" to "how did you get there." Once the process stays visible, it becomes clear how far a student actually thought, whether they used AI or not.

Designing assessment that keeps the process, with Flipsson

Flipsson, Nallijaku's modular lesson platform, is built so that making, teaching, and checking continue in one place, which suits process-preserving assessment especially well. Open it in the browser with nothing to install and you can set it up like this.

Turn on a live lesson and students write their answers in class, within a set time. They come in with a join code and write right there with screens in sync, so what stays in the classroom isn't a product finished at home but the thinking of this very moment. Split the written-response blocks in the block editor into several parts and take them in stages — "evidence → claim → counterargument" — and instead of a one-line conclusion, the grain of thinking as it grows shows through. The way a student changes their mind piles up as cards on the opinion and team board, and even the traces of reviewing each other's cards through comments stay behind.

You can also tame AI rather than shut it out. The AI Tutor answers from the module being studied rather than the whole internet and shows which material each answer came from, so what a student asked and what they consulted is visible. AI becomes a tool for asking about where you're stuck, not a tool that writes the answer for you. The teacher scans the flow of participation and submissions in the dashboard and designs the next round of feedback on the basis of how students got there, not what they turned in.

When the process is visible, AI stops being a threat and becomes a partner in thinking. This is exactly where assessment turns from surveillance into watching growth.

Start with one question

You don't need to overhaul your whole assessment system at once. Take one question from your next performance task, collect it as a staged written response inside a live lesson, and watch a student's thinking pile up as cards. Flipsson starts with no card on file and is free for up to 10 classes, so piloting it in a single class costs you nothing. Once the process starts staying behind, the question "did you really write this?" stops being necessary at all.