AI & Academic Integrity: Design, Don't Detect

Move beyond detector-led responses by clarifying AI expectations, collecting evidence of process and designing tasks that make learner thinking visible.

• Design before you detect — Treat integrity as a learning-design problem before it becomes an enforcement problem.
• Define what AI use means for this task — Replace generic 'AI allowed' or 'AI banned' statements with task-specific boundaries.
• Collect evidence while learning is happening — Use drafts, conferences, annotations and decision trails to see the learner behind the product.
• Use conversation when evidence does not align — Treat inconsistency as a prompt for inquiry rather than an automatic accusation.
• Build a culture of disclosure and judgement — Make appropriate AI use discussable so students do not learn that all use must be hidden.