Professional learning · Teachers, Coaches · 60 minutes
Thinking AI Can't Fake
Teachers and coaches take an assignment a chatbot can finish in ten seconds and redesign it so the thinking stays with the student, trading plagiarism panic for tasks that make reasoning visible and resist cognitive offloading.
Most AI-in-school anxiety is really about assignments that AI can now complete invisibly, which turns teachers into detectives and students into suspects. This session offers a better move: stop policing the tool and redesign the task so the thinking is the point and the evidence. Participants take a real assignment they already use, test how easily a chatbot could do it, and then rebuild it using moves that resist cognitive offloading, requiring in-process reasoning, personal or local specifics, defense of choices, revision based on feedback, and comparison or critique of an AI answer rather than acceptance of it. Coaches practice the cycle (listen, co-plan, model, measure, adjust) around a teacher's redesign and name the evidence to watch. The stance is balanced: AI is not forbidden in the redesigned task, often it is deliberately part of it, but the student's thinking is what gets made visible and assessed. The guiding question stays the ELE one: what evidence shows the learning, with or without the tool?
In this packet
- Handout A: The AI Stress Test (worksheet)
- Handout B: Seven Redesign Moves (worksheet)
- Handout C: Coaching the Redesign (worksheet)
TCEA ELE indicators
- T5.1 I know how to design lessons that promote higher-order thinking skills.
- T2.1 I can teach students to critically evaluate online information.
- C2.1 I know how to guide teachers in meaningful technology integration.
- AI4.1 I know how to promote critical thinking skills in relation to AI-generated information.
Full facilitator guide: mglearn.github.io/eles/activities/thinking-ai-cant-fake.html
Handout A for Thinking AI Can't Fake
The AI Stress Test
Run the assignment you brought through this test. Be honest about how exposed it is, that's the opening for a better task.
What does the assignment ask students to turn in, and does it reward the PRODUCT or the THINKING behind it?
On a scale of 1–5, how completely could a chatbot do this with no trace? What exactly could it produce?
What thinking was this assignment SUPPOSED to grow? Does the current version actually require it?
The gap: where does a slick AI answer fall short of real understanding on this task?
Handout B for Thinking AI Can't Fake
Seven Redesign Moves
Pick at least TWO moves and rebuild your task below. The goal is a better task, not just a harder-to-cheat one. Several moves use AI on purpose, with the student's judgment as the graded part.
THE MOVES (reference): 1) Require in-process evidence (drafts, reasoning logs). 2) Anchor in personal or local specifics AI can't know. 3) Require students to defend their choices, out loud or in writing. 4) Build in revision after feedback. 5) Have students compare or critique an AI answer rather than produce one. 6) Shift key reasoning to in-class performance. 7) Assess the process, not just the product. Which two (or more) will you use?
MY REDESIGNED TASK (write the new version students would receive):
What thinking does this now require that AI can't fake?
What evidence of learning will I actually see from this task?
Handout C for Thinking AI Can't Fake
Coaching the Redesign
Frame a coaching cycle around a teacher's redesign. Keep every step anchored to student evidence, not tool detection.
LISTEN: What does the teacher want students to be able to do? What's the worry behind the assignment?
CO-PLAN: Which redesign moves, and what evidence of learning will we look for together?
MODEL / MEASURE: How and when will we examine the resulting student work side by side?
ADJUST: Our coaching question is "What changed in student evidence?" What will we change next cycle?