PN20: Assessment – The Open Lane

Much of what we assess already sits in the open lane, labelled or not: anything a student completes out of sight, over days, is open in practice. Calling a task open settles the permission question and leaves the design question untouched. Corbin, Dawson and Liu (2025) argue that open assessments stand or fall on their structural design, and that ever more detailed rules about acceptable AI use only widen the gap between what we can specify and what we can verify.

I would argue the open lane is unusually well placed to assess one learning outcome: evaluative judgement, the capability to judge the quality of work (Bearman et al., 2024). An AI tool can produce the work. Deciding whether it is any good, and defending that decision, remains the student’s job. The bounded middle lane needs the same structural thinking, and Curtis (2025) reminds us that if we expect students to use these tools, we have to show them how.

A public health vignette shows the shift in practice: an AI-drafted policy brief, annotated and revised by the student, with half the marks on the annotation and five minutes of explanation in a seminar.