Lydia Arnold

In recent talks that I have given about artificial intelligence in learning and assessment, one point has piqued quite a bit of interest. In short – I have argued that pride and joy should be a North Star in assessment design.

Academic conversations around AI overwhelmingly focus on assessment integrity. That’s important, but I’m concerned that this focus comes at the expense of other issues. Amongst this hyperfocus we risk losing sight of the intrinsic joy and satisfaction that can be brought about by learning and producing artefacts for assessment. Within the assessment debate, when we talk about cheating and poor behaviour, there’s often an assumption that AI will be used as a shortcut or in a problematic way. It offers a hole in the fence through which students will choose to run. There is no denying that this is happening, but that is only one aspect of the assessment landscape. 

John Warner said that ChatGPT won’t kill anything worth preserving. I agree.  When I speak to students in recent weeks about their dissertations, they’re full of pride and curiosity: “How can I find out the answers to these questions? How can I explore this topic?” Learning is a process full of fizz and joy, wide eyes and possibilities. The risk is that AI steals our joy, creates distrust, and makes us forget that assessment can and should be for learning. It should be energising, joyful and motivating. For some this may be about imaginative tasks, for others it may come through mastery of mathematical processes or recreating an experiment that worked well. Whilst we won’t enjoy all we learn or create or achieve, this shouldn’t take away from orienting learning toward joy and intrinsic motivation. I’m not arguing that assessment joy will stop ‘cheating’ but perhaps it will help a little if we can design assessments, and support learning, in ways that foster pride and personal investment, so that we talk less about security. 

Taking pride in assessment doesn’t mean avoiding AI use. Sometimes it may be that AI isn’t used, but other times it may be a learning tool – one example is working on complex data and spreadsheet tasks. AI can help with learning, testing, failing, iterating, refining, and developing an understanding of which formulas work and which do not. 

How do we design assessments that achieve this and shift the dialogue from integrity toward motivation? The answer is likely multi-faceted but may include recognising learner individuality and different ways to shine, and no doubt it will involve scaffolding and encouraging mastery. As a sector we might also reflect on our assessment infrastructure (use of criteria, anonymous submissions) and assumptions (test every learning outcome), asking whether they serve us well. There may also be benefit in looking at other areas of education. I have long thought primary schools are places of joyful learning for many (not all, I get that). Primary level joy is a researched area, with advice including “[t]he joy of learning does not like to hurry” and “[t]he joy of learning does not include listening to prolonged speeches” (Rantala, T., & Määttä, 2012). I can’t help feeling primary education is a place to learn from in a world of GenAI.  

Rantala, T., & Määttä, K. (2012). Ten theses of the joy of learning at primary schools. Early Child Development and Care, 182(1), 87–105. https://doi.org10.1080/03004430.2010.545124

Warner, J, (2022) ChatGPT Can’t Kill Anything Worth Preserving. https://biblioracle.substack.com/p/chatgpt-cant-kill-anything-worth

Copilot used as a minor editing partner.