Assessment integrity and validity in the teaching laboratory: adapting to GenAI by developing an understanding of the verifiable learning objectives behind laboratory assessment selection

Abstract

Generative Artificial Intelligence (GenAI), such as ChatGPT, is reshaping educational paradigms by offering unparalleled benefits and introducing challenges, particularly academic integrity. This study investigates teaching laboratory practices (traditional, recorded, remote, simulation and virtual), considered an academic safe haven due to its authenticity, and examines how assessments align with learning objectives. This should reinvigorate interest in expanding laboratory learning opportunities. However, unsupervised laboratory reports, a dominant assessment type, present significant cheating risks–intensified by GenAI. Given the scant literature on laboratory assessments and their primary focus on cognitive objectives, little guidance is available regarding how to assess non-cognitive objectives. This studies innovative approach utilises a reflective survey with 134 international academic staff to explore how each assessment type can verify cognitive, psychomotor, and affective learning objectives. We introduce a ‘Words of Estimative Probability’ heatmap to visualise the likelihood of verifying specific learning objectives, providing a snapshot to guide academics in holistic assessment design. This study advocates for diverse assessments, which mitigate GenAI risks and foster comprehensive skill development. This research equips educators to design secure, effective laboratory education in STEM disciplines, ensuring alignment with evolving academic and technological landscapes by offering a framework for improving assessment validity, integrity, and adaptability.