Fostering Student Agency & Leveraging AI

Students in the same class can follow different assignment routes. One option is to offer a menu of tasks with different marks, formats, and levels of difficulty. A student might combine revision or recall work with writing, programming, a demonstration, research, teamwork, or an open-ended problem that draws on several concepts. Everyone works towards the same total, and students control part of the route.

Choice needs clear boundaries. Publish criteria for each option, including format-specific criteria for programming and research tasks, and assess the evidence of learning each task is designed to produce. The assignment should help students learn and show how they learnt. Start small: offer two or three options in one course, track which ones students choose and how much marking they create, then adjust the menu next semester.

The same approach can make generative AI part of the assessed work. Students first solve a reference problem independently. They then give it to three AI systems, compare those responses with an authoritative solution, revise their prompts, and record the tools, criteria, discrepancies, and final recommendation. Treat agreement between several tools as a finding to verify with disciplinary knowledge and evidence.

Assessment can focus on judgement: Which answer is accurate? Which explanation is clearer? What assumptions or errors appear? When is one method more efficient? Require students to explain how they handled conflicting results. Marks should support the learning task. A sound assignment gives students a choice within clear limits and makes them show how they reached a decision.