Contextual Background
Most core users of the Grow Lab have no background in science or laboratory practices. This is not, an impairment for anyone to work in the lab, for we provide inductions, training and support continuously. However, the use of AI to complete planning forms, generate protocols or come up with project ideas, frequently undermines my ability to assess students’ actual understanding. This creates a level of distrust, for it can be a safety concern.
Evaluation
Technician assessment is typically informal and dynamic than traditional academic ones. Although we have a Grow Lab form that acts as scaffolding (Vygotsky 1978) for student organization prior to lab sessions, this can be completed using AI.
Since our primary assessments happen through face-to-face interaction with students in the lab, we can get information from them directly, and assess AI influence. We attempt to do what Wiggins (1990) calls authentic assessment by observing their confidence, language, and experimental plans. We are assessing for their theoretical domain of the subject through their practice, but also, for tacit knowledge (Polanyi 1966).
AI has limitations and does not take our specific context in consideration. When a proposed work is not compatible with the student’s comprehension of the topic, we suspect of AI use. The more a student masters issues related to their project, practical and theoretical, the more they will develop autonomy in the lab. Their independence signals the successful learning and the effectiveness of our approach.
Moving forwards
Our main strategy must be to be constantly aware, to question without intimidating, continuously ask for their sources, and assess their understanding, while allowing progressively more independence. If we suspect of AI use, we ask for an original non-AI source, go through the original finding versus the AI response, discussing its limitations and possible safety issues.
Moving forward, I will look into creating tools guiding students through good AI use for laboratory practices, giving examples of health and safety and intellectual property concerns and highlighting UAL guidelines on AI (UAL, Nd). I have been approaching the topic through staff development workshops on AI and academic integrity and contacting colleagues from other laboratory institutions (such as RCA) to discuss how they approach the subject. One common strategy is to have “checkpoint questions” during consultations and conversations like “walk me through your plan” and “what happens if this fails”. We currently do it intuitively during conversations but might be a good idea to present them more frequently as provocations for critical reflection – in the Grow Lab form, inductions or other tools.
We cannot provide the intensive support that we would like as technicians, which would allow on our part also, a better understanding of student’s comprehension, limitations, needs and a more tailored support to each. The solution, as suggested by Rowe and Potier (2026) in their AI workshop, is that students get more “time in the physical world”, or in this case, more time in the lab environment. If we cannot provide them with more time, we need to work on providing the best quality in the time they have.
References
Wiggins, G. (1990) ‘The case for authentic assessment’, Practical Assessment, Research, and Evaluation, 2(1), Article 2. Available at: https://doi.org/10.7275/ffb1-mm19
Vygotsky, L. S. (1978) Mind in Society: The Development of Higher Psychological Processes. Cambridge, MA: Harvard University Press.
University of the Arts London (no date) AI and Arts Education. Available at: https://www.arts.ac.uk/about-ual/learning-and-teaching/digital-learning/ai-and-education (Accessed: 22 March 2026).
Rowe, C. and Potier, R. (2026) ‘Art, Design and Artificial Intelligence’ [Guest lecture]. PGCert in Academic Practice, University of the Arts London, 19 March.
Polanyi, M. (1966) The Tacit Dimension. London: Routledge & Kegan Paul (cited in Cleary, V., 2026).