- Video Insight
A New Way Forward – Redesigning Assessment in the Age of AI
- 2 mins
FULL TRANSCRIPT
Alex Fuller and Myles Aravindhan
Rethink a new way forward. So we’re talking about redesign, not just defend here. Design with AI in mind, not against it. What’s that mean for us, Miles?
Well, this kind of goes back to that original slido — that confidence, and some people saying they don’t want to think about it. I know it’s a funny response, but in a way your assessment will echo that if you build it hoping that AI isn’t being used.
So what this one means is that whenever we write a question, whenever we write a task, we should assume that AI is available to the student. So if AI is available to the student, then we ought to avoid questions that AI can answer.
So what that means is let’s avoid superficial questions and let’s dig a little deeper. Because if we ask a question that AI can answer — the way I always saw it — if AI can answer it, then there’s no point asking that question to begin with.
So let’s move away from that idea of surface-level “what is” or “identify” questions. Instead, let’s ask questions that require a student to apply knowledge to their own specific context — their own life, their own workplace, their own experience — because that’s something AI can’t replicate for them.
And again, this kind of adds to that overarching idea of multiple processes at once — tasks that require a bit more depth, that pull the question away from AI by another step, and move it back to the student by another step.
Key Points
- Design with AI in mind, not against it. Building assessments that simply hope students won't use AI is a losing strategy; assessments should be architected assuming AI is available.
- Assume AI is available to every student. This mindset shift forces educators to write questions that AI alone cannot satisfactorily answer.
- Avoid superficial "what is" and "identify" questions. These can be resolved instantly by AI tools and therefore offer little evidence of a student's own capability.
- Go deeper and more personal. Questions that ask students to connect concepts to their own experience, workplace, or specific context are much harder for AI to answer authentically on their behalf.
- Prioritise multi-step and process-based tasks. Assessments that require students to demonstrate layered thinking — multiple processes at once — move evaluation further from what AI can replicate and closer to genuine student competence.
- The redesign mindset is an ongoing process. Rather than a one-time fix, this approach involves continually revisiting and refining questions to keep pace with evolving AI capabilities.