Topic 4 brought together threads I had encountered separately before but rarely seen woven together in one conversation: how we design learning, how we assess it honestly, how we build the conditions for trust and inclusion, and what AI now means for all of it. The combination made for one of the more thought-provoking sessions in ONL for me.
Backwards, but in a good way
In out group we talked about the Understanding by Design framework which was not new to me. I had come across Wiggins and McTighe’s backwards design logic before, start with the desired outcomes, then design assessments, then plan learning activities. But revisiting it in this context was a useful reminder of how easy it is to drift back to forward planning in practice: coverage first, assessment almost as an afterthought. What the framework insists on is discipline, the clarity to ask what does genuine understanding look like here? before deciding what to teach or how to teach it. That question sounds simple but it is surprisingly difficult to answer well, especially in higher education where content coverage often feels like the point rather than the means.
What resonated more this time was the emphasis on transfer, whether students can take what they have learned and apply it somewhere new, without being prompted. That is a much higher bar than recognition or reproduction, and it quietly reframes assessment from a measurement exercise into a design challenge.
Assessment that holds up
One of our teammates had a presentation on AI-resistant assessment and we decided to take that and review it. It actually landed at exactly the right moment. I went in slightly skeptical, the phrase “AI-resistant” risks sounding like we are simply making things harder for students rather than making assessment more meaningful. But the argument was more sophisticated than that. The assessments most likely to survive AI are the ones that were never well-designed to begin with: decontextualised essays, generic reflections, tasks with no stake in the real world. Designing assessments that require students to engage with their specific context, their own data, their own position, or a live and unpredictable situation is not a workaround, it is just better assessment design.
That framing helped me. Rather than thinking about how to police AI use, the more productive question is: would this assessment still be meaningful if a student used AI to complete it? If the answer is yes, the problem is not AI, the problem is the assessment.
Trust and inclusion as design decisions
One idea that stayed with me from this topic is that trust and inclusion are not soft add-ons to learning design, they are structural. Students will not take intellectual risks, engage in genuine dialogue, or produce honest reflections in environments where they do not feel safe or seen. In a university context, that is easy to underestimate. We tend to assume that adult learners arrive ready to engage, but readiness is something that has to be designed for and maintained, not assumed.
This connects directly to my ONL group experience. The collaboration took time to warm up, and in retrospect a lot of that was about trust building, learning who would take initiative, whose voice carried weight, whether it was safe to disagree or be uncertain. If that is true among a small group of motivated professionals on a course we chose to join, it is certainly true for students navigating a compulsory module at a university.