What you will learn
- Explain why AI detectors cannot be the main line of defence.
- Redesign assessment so that outsourcing the work defeats the point.
- Build an integrity approach that protects students rather than surveils them.
Integrity is protected by the design of the task, not by surveillance. This module replaces the detector arms race with assessment that AI cannot quietly complete.
Why detectors are the wrong foundation
Start by accepting that AI detectors do not work reliably. They miss lightly edited AI text, and worse, they falsely accuse real students. The false accusations fall hardest on students who write in a second language, whose plainer phrasing reads as machine-like to a detector. A school that disciplines a child on a detector's say-so will eventually be wrong about a real person, and that single failure costs more trust than the policy was ever worth.
Redesign the tasks that break
The durable defence is assessment design. Use your classroom pilots to find which tasks a chatbot can now complete in seconds, and change them.
| Fragile task | Redesigned task |
|---|---|
| Take-home essay on a common topic | In-class writing, plus an oral defence of the argument |
| Summarise this chapter | Critique an AI summary of the chapter and correct its errors |
| Answer these ten questions at home | Hand in drafts, notes, and a short reflection on your process |
| Write a report on a country | Compare two AI-generated reports and judge which is more reliable, with evidence |
Try this
Take one assignment you already set. Feed it to a chatbot yourself. If the output would earn a passing grade with no student thinking involved, the task needs redesigning, not policing. Rework it using one row of the table above.
Design so cheating and learning cost the same
The principle underneath all of this: when the honest path and the shortcut require the same effort, most students choose to learn. When outsourcing is far easier than thinking, some will outsource, and no detector will reliably catch them. Move the effort to where the tool cannot follow, in-class work, oral defence, process artefacts, and integrity mostly takes care of itself.
What the frameworks say
TeachAI's toolkit treats academic integrity as a design and policy question, offering adaptable language on responsible versus prohibited use. The Australian Framework's Accountability and Transparency principles ask schools to make expectations for AI use explicit rather than to rely on detection.
- TeachAI: AI Guidance for Schools Toolkit (TeachAI (Code.org, ETS, ISTE, Khan Academy, World Economic Forum))
- Australian Framework for Generative AI in Schools (Australian Government, Department of Education (2023))
Sources for this module
- TeachAI: AI Guidance for Schools Toolkit, TeachAI (Code.org, ETS, ISTE, Khan Academy, World Economic Forum). https://www.teachai.org
- Australian Framework for Generative AI in Schools, Australian Government, Department of Education (2023). https://www.education.gov.au/schooling/resources/australian-framework-generative-artificial-intelligence-ai-schools