Designing AI-Resistant Assignments: A Practical Framework

Designing AI-Resistant Assignments: A Practical Framework

← The Journal

Every teacher has felt it by now: the essay that reads a little too smoothly, the problem set finished a little too fast. The instinct is to reach for detection tools. The better move is to redesign the task itself.

The one-question audit

Before assigning anything, ask: could a student complete this by asking an AI and copying the answer? If yes, the task is measuring access to a tool, not understanding.

The strongest assignments require judgment, presence, and revision — three things a chatbot cannot deliver on a student's behalf.

Four moves that work

  • Anchor in local evidence. Use sources from your classroom, your community, or your students' own data collection.
  • Require positions to change. Ask students to revise their argument after new evidence is introduced mid-unit.
  • Make thinking public. Oral defenses, seminars, and exhibitions put understanding on display in real time.
  • Assess the process. Grade drafts, annotations, and reflections — not just the polished artifact.

None of this requires banning technology. It requires designing work where the thinking is the point — and where doing the thinking is the only way through.

Your next great unit starts with a question.