When students can use AI to generate code, explain syntax, and accelerate routine development tasks, the challenge for teachers is no longer just how to assess coding. It is how to keep students genuinely engaged, how to value the design, judgment, communication, and reflection that sit around a software project, and how to structure complex project work so that students can succeed rather than stall at the end. In my dashboarding unit, these pressures came together in a large, project-based subject where students had to build a substantial data product over the semester.
This talk shares how I responded by redesigning the unit around a feed-forward assessment model. Instead of relying mainly on post-submission feedback, I broke the project into authentic stages, replaced lectures before each deadline with draft workshops, built in peer evaluation that assessed the quality of students’ critique rather than their peers’ marks, and used a final showcase to make student work visible to the wider cohort. The result was an assessment design that focused less on what students could produce at the last minute and more on how they developed ideas, responded to feedback, collaborated, and improved over time.
For educators teaching Python, data science, or project-based computing subjects, this offers a practical way to design assessment for an AI-open classroom without falling back on bans or surveillance.
Stephen is a Senior Lecturer at the University of Sydney in the fields of Statistics, Data Science and Machine Learning.
Stephen co-founded Wattle Education, with Alison Wong, which produces resources for Stage 6 Software Engineering teachers in NSW.
