Free. Five minutes. One assignment.
Could a chatbot pass your assignment?
The Assignment Audit finds the lowest level at which a student could complete your task and still earn full marks, then names the one thing worth changing first.
The test that starts it
Take an assignment you're using this term or semester. Paste the instructions into a chatbot exactly as your students receive them. Read what comes back and ask yourself one question: would you give this a passing grade? If you would, the assignment needs redesigning. Knowing this is the whole point of the exercise.
Most assignments fail this test the first time, including assignments that were excellent in 2019. They were designed against a different problem.
What AI can and cannot do in your classroom
| AI can do this easily | AI cannot authentically do this |
|---|---|
| Write a polished five-paragraph essay on any text | Reference a class discussion from last Tuesday |
| Produce sophisticated vocabulary and varied syntax | Show a genuinely messy first draft with crossed-out ideas |
| Generate plausible-sounding textual evidence | Explain why a thesis changed between Draft 1 and Draft 2 |
| Mimic a student's established writing style | Write in real time, under a time constraint, while you watch |
| Create multiple versions of any essay on demand | Respond to specific peer feedback in the student's own voice |
| Pass most AI detection tools | Demonstrate growth across three visible, dated drafts |
Everything in the right-hand column is evidence you can design an assignment to collect. That's the whole method.
The Five-step Redesign
Once an assignment has failed the test, this is the procedure. It works on essays, research tasks, creative writing, and reading responses.
- Run the AI test
Paste the instructions in, read the response, and decide whether it would pass. If it would, carry on to Step 2.
- Map it against the four pillars
Ask four questions. Does the assignment collect evidence of the student's thinking process, not just the final text? Does any part of it happen in the classroom in front of you? Are the AI expectations clear enough that a student knows what is permitted? Does it build the student's understanding of what AI is good and bad at?
- Add the missing pillar
Most assignments fail on the first or the second pillar, because there is no process work or because everything happens at home. Work on the weakest pillar first. Missing process means adding stages: brainstorm, annotated draft, peer review, and reflection. Missing in-class cognition means adding a timed write, an exit conversation, an oral defence, or a cold reading. Missing clarity means adding a permission label. Missing meta-literacy means asking students to evaluate an AI response to the same prompt.
- Add the spotlight sentence
In the final stage, every student identifies one sentence or idea in their work that only they could have written and explains why. It takes about thirty seconds per student, and it reveals more about authenticity than any detection software.
- Run the test again
Paste the redesigned assignment back in. The chatbot response should now fail, because it cannot supply the observed process, the in-class component, the personal specificity, or the declared AI interaction. If it still passes, return to Step 3.
Send me the audit
Free, and it arrives in about a minute.
- The Assignment Audit, six pages, print and go
- The interactive version that scores it for you and names your weakest answers
- The IGNITE Framework poster in editable PowerPoint and vector, so you can put your own department name on it
An occasional email, usually every week or two, about assessment that holds up when students have AI. Unsubscribe in one click, any time. I will not share your address with anyone.
Where it fits in the framework
The audit is the diagnostic. It tells you which IGNITE level your assignment is currently asking for, which is often lower than the level you thought you had set. The framework then gives you the vocabulary to say what you want instead, in words your department will recognise.
Questions teachers ask
How long does the audit take?
About five minutes for one assignment, once you have the assignment in front of you. The redesign takes longer, but you'll usually find that one change carries most of the weight.
Do I have to redesign everything I teach?
No, and you shouldn't try to. Start with the assignment that worries you most, usually it's the major take-home essay. Fixing the weakest pillar on one task teaches you the move, and the move transfers.
What if my assignment fails the test badly?
Most do, including good ones written before 2023. A task can be pedagogically sound and still be completable by a chatbot, because it was designed against a different problem. That's a design gap, not a judgement on the teaching.
Is this only for essays?
The procedure works on essays, research tasks, creative writing, and reading responses: anything where a student hands you a finished text they produced somewhere you didn't see them writing.
The framework behind the audit
Six levels, four pillars, and the weighting that makes the whole thing hold together in a real course.
Read the frameworkBrowse the resources