Designing for It
Designing assignments that demonstrate
what AI skips in the learning process
Students that use AI to bypass cognitive work don’t often realize that they are skipping over an essential component of their education. This entry will help faculty identify the important cognitive work done during the assignment, and then provides suggestions for how to make that explicit and clear to the students.
The concept
Students develop stronger skills from doing rather than observing. By participating in the frition of cognitive work (making decisions, dealing with uncertainty, correcting themselves, etc.) students undergo the process of learning.
In the case where AI produces the work with minimal input and interaction from the student, a product is outputed but learning is bypassed. For busy students that are most concerned with submitting a product that leads to a high grade, using AI is tempting, but students can’t see what they are missing. These students rarely have reference points for what cognitive work feels like, and what using AI in this way is actually costing them in the long run.
The first prompt in this entry helps faculty identify what a student is missing by allowing AI to do all the work for them. The follow up prompt helps develop a metacognitive component to the assignment that allows students to reflect on the work they did and recognize what would have been lost if AI did the work for them.
The diagnostic
Two minutes before you prompt-
1What skill is this assignment actually developing?
Ignoring the final product, why are students doing this activity? Do you want them to practice synthesizing different sources? Do you want them to form a judgement form ambiguous evidence? Do you want them to construct a position and defend it?
-
2If AI completed this assignment, what would students have not done (be specific)?
Remember to be honest here. Would the cognitive work you named above been entirely skipped or would students still need to do some of it?
-
3After completing the assignment, could your students clearly describe what cognitive work felt like and what skills they built?
Have you built in enough structure for students to come to the conclusions you’ve set out in your assignment? Are you asking for just a product without any space to reflect on the learning process your students have just experienced?
-
4What could you modify to make this learning more visible to students?
This could be as simple as a short reflection, discussion question, or self-assessment tied to specific cognitive work. This is your chance to focus their attention on what they learned, not just what they produced.
The prompt
Here is an assignment I use:
[Paste your full assignment prompt here]
Show me in concrete detail how a student could use AI to complete this assignment. Include the specific prompts they might type and what the output would look like. Don’t hold back — I want to understand the full extent of what AI could do here. — Step 2: Design the metacognitive component Now I want to build metacognitive awareness into this assignment. The goal is for students to be able to reflect — specifically and concretely — on the cognitive work they did, and to recognize what they developed that AI couldn’t have developed for them.
This assignment is meant to develop [your learning goal, e.g. “the ability to analyze primary sources and form an original historical argument”].
Suggest 2–3 specific prompts, questions, or structured moments I could add to this assignment that would make this learning visible to students. These shouldn’t be generic reflection questions — they should be tied directly to the cognitive work you identified above. I want something a student who did the work could answer in depth, and a student who used AI to skip it would struggle to answer at all.
↑ Run step 1 first, then follow up with step 2 in the same conversation so the AI has context
Write a prompt for your assignment that addresses the metacognition behind the assignment. Keep it short and concise, and try to tie it directly to the cognitive work you think this assignment should develop.
Instead of asking students what they think they’ve learned, you might start the assignment with a short explanation of what skills they’ll develop by doing this work and why it matters (especially when AI could do some of the work for them).
Make a rubric to assess the metacognitive component and attach it to the assignment. Make sure students know what differs between a strong response (a student that learned something) and a weak response (a student that didn’t learn as much).
Prompt variables
What this looks like in practice
“Write a 1,000-word essay analyzing the rhetorical strategies in this speech and arguing which was most effective. Cite at least three moments from the text.”
A student could complete this with AI without reading the speech and produce an essay structurally indistinguishable from genuine analysis.
Same essay. Within 24 hours of submitting, answer this in 150 words:
“Which rhetorical strategy did you almost argue was most effective — but didn’t? Walk me through what you found in the text that made you move away from it.”
What responses reveal
“I originally planned to argue ethos — the speaker’s references to personal experience felt like the spine of the speech. But in the third section, the emotional appeals were doing something different: they were asking the audience to feel responsible, not just informed. That shift is what changed my argument.”
“I considered all three strategies and ultimately decided pathos was most effective because it had the strongest emotional impact on the audience and was used consistently throughout the speech.”
Anticipated results
The first part of this entry produces a description of what things AI can “replace” in your students’ learning journey.
The rest of the prompt will help you think through how you can add metacognition to your assignment. Expect to refine it, but the first draft is usually close.
If you do the “explain it” followup, you’ll have drafted a few sentences that you can provide students before the start of the assignment so that they are aware of why they are doing the assignment and what they can plan on learning/gaining.
Caveats
- Students might also use AI to complete the metacognitive part of the assignment as well. The metacognition component is not foolproof, but one could argue that even using AI to perform metacognition requires some interaction with the material. If you are really concerned with this, consider pairing the assignment with an in-class followup where students must argue their stances with each other.
- The more specific you are with the metacognition, the more everyone will get out of it. Try to avoid generic questions such as “what did you learn” or “how did the assignment go for you”. These types of questions are easy to respond to without having done the difficult cognitive work.
Reflection questions
- What would it look like for a student to walk out of this assignment with language for what they developed — not just a product they submitted?
- If you shared the metacognitive prompt with students before they started the assignment rather than after, how might it change how they approached the work?
- What would it mean to teach this diagnostic process explicitly — to walk students through it on one of their own assignments and let them identify what AI would skip for them?
