Step 4 of 5 · The method as a teaching tool

Bringing the method into your teaching

You can also build the three-step protocol into the way your students work with AI. This page shows how.

Protect the moments where learning happens. Let students use AI for tasks that do not require them to think, and keep the thinking that builds lasting knowledge as their own work.

The structure: before, during, after

You can design any assignment around these three stages, as long as you make them explicit. Students should know what they are doing at each point, and why.

Before · the student works first

Activate prior thinking before AI is opened

Ask students to write down what they already know, believe or are unsure about, before they touch any AI tool. A few sentences is enough. It gets their thinking down where they can see it, and gives them something to compare with after the AI conversation.

Example prompt for students: “Before you open any AI tool, write two or three sentences: what do you already think about this question? What are you unsure of?”

During · structured engagement

Build in a moment of metacognitive pause

At some point during the AI interaction, students stop and ask themselves whether the AI is helping them think or thinking for them. It is a small step with a large effect, because it interrupts passive consumption and makes students engage with what they are reading.

The pause question“Is AI helping me think right now, or is it thinking for me?”

After · close the loop

Require reflection on what was learned and how

Students return to what they wrote before the AI conversation and compare it to where they ended up. What changed? What did AI add? What do they still need to work out? Done regularly, this closes the loop and builds a metacognitive habit.

What did AI say, and what matched or didn't match your own thinking?
What surprised you?
What would you still need to work out without AI?

How it looks in different disciplines

The structure is the same across disciplines: only the content of the task changes.

Health sciences

Task: analysing a clinical case. Students write their own differential diagnosis before consulting AI. During: pause and check whether they are evaluating AI's reasoning or accepting it. After: what would they change, and why?

Humanities and social sciences

Task: developing an argument. Students outline their thesis in their own words first. During: does AI flatten the complexity, or introduce perspectives they hadn't considered? After: is the argument still theirs?

Engineering and STEM

Task: solving a problem. Students attempt the problem, or part of it, before consulting AI. During: are they understanding each step, or copying output? After: could they solve a similar problem now without AI?

Any discipline

Task: any written assignment. Students submit a short “AI use log” alongside their work: what they used AI for, what they kept, what they changed, and what they learned from the process.

A practical note

You do not need to redesign an entire course to try this. The simplest version is to add one sentence to an existing assignment:

“Before opening any AI tool, write a short note on what you already think. Submit this alongside your work.”

Writing down prior thinking before AI is enough to interrupt the most passive forms of cognitive offloading. Start with that, see what happens, and build from there.

What this approach does not do

The method keeps the chance to learn open, but it guarantees nothing, and students who are set on skipping the before-step will skip it. What you are after is a shared understanding with your students of why the struggle matters, and what they lose when they skip it. Checking up on them will not get you there.

One last step: your commitment. Move to Step 5 and note one concrete situation where you will try this with your students.