Abstract
The design of effective assessment has long presented a complex challenge for educators, one that has intensified with the emergence of Generative Artificial Intelligence (GenAI). The maintenance of assessment validity in the context of GenAI has been characterised in the literature as a wicked problem: a multifaceted issue lacking a singular solution and necessitating structural rather than purely discursive responses. This paper advances a blueprint for one such structural response premised upon evaluating the processes through which artefacts are produced. We engage the concept of Black Box Thinking to propose a processual assessment design model termed Black Box Assessment. This model is underpinned by four core principles: (1) assessing process to capture evidence of learning and cognitive change; (2) valuing errors and revisions as indicators of progress; (3) fostering transparency and learner agency; and (4) rewarding the learning trajectory not just polished products. We outline the key steps in designing a Black Box Assessment and tackle potential challenges to this approach. We conclude by arguing that Black Box Thinking provides not only a conceptual basis for process-focused assessment design but also a valuable evaluative framework for addressing the wicked problem of assessment in the age of GenAI.