Investigating Web Project Assessment in an AI World

Project-based assessment in university web application courses is widely used because it shows how students integrate skills and knowledge across development. However, in large cohorts, project assessment becomes time-intensive and it is difficult to keep evaluation consistent across students. Automated support can reduce workload, but may limit instructors' agency during the assessment. To understand instructors' challenges and needs, we conducted semi-structured interviews followed by a sketching exercise with seven university instructors who regularly assess web application projects as part of the final exam. By analyzing data from both activities, we derived four themes and related implications. Key themes include critical thinking as the boundary for acceptable GenAI use and instructors' preference for instructor-in-the-loop tools that automate pre-checks and extract evidence to drive oral questioning. These findings can inform the development of systems that can support instructors in assessing web application projects at scale while keeping them in control of decisions.