Our Head of Assessment Training, Dr Simon Child recently attended the annual Assessment in Higher Education conference (AHE) in Manchester. This event brought together HE specialists from across the globe to explore and navigate the evolving landscape of assessments in higher education.
In this blog post, we'll delve into some of Simon's reactions from the conference, focusing on threats and possibilities that developments in could AI bring, and how approaches such as authentic assessment can be considered to mitigate against the challenges faced by the sector.
"I was fortunate to attend the Assessment in Higher Education conference in Manchester in late June, where some of the latest developments in Higher Education assessment were discussed.
Perhaps unsurprisingly, the big themes of the conference were authentic assessment, inclusivity, and artificial intelligence - with the majority of the talks addressing one or more of these topics.
The conference was certainly useful for people trying to navigate the disruption caused by AI. In their keynote ‘Assessment reform in the age of Artificial intelligence’, Phillip Dawson and Thomas Corbin argued that we are now able to offload cognitive effort onto AI, and because traditional assessments achieve task validity by representing this cognitive effort, there presents a risk to academic integrity.
Indeed, a recent research report by the University of Reading found that AI can perform better than actual students in modular exams, particularly in earlier years of the course programme.
On the face of it, this is a worrying prospect and there was a clear undercurrent of concern amongst academic leaders at the conference. While some universities have responded by changing the security parameters around their assessments (for example by re-introducing in-person examinations) Dawson and Corbin argued that there is an opportunity to ‘lean in’ to the reality of AI, by creating assessments that focus less on the output of cognitive endeavour (such as an essay) but rather focus on the process of creating such outputs.
One example they gave was to provide an AI-generated argument related to a subject discipline, and asking students to critique the qualities of the arguments made. They reported on research that indicates AI is poor at these higher-order tasks, and so at least in the short term, assessments that focus on these critical skills are authentic to what students will likely be doing after graduation.
Similarly, James Mapp at the University of East Anglia presented an example where students were asked to discuss the decisions made on their computer code in a short viva, with an expectation that they will use AI to generate at least some of the code. What was so interesting about this research was the overarching concern with both lecturer workload and student well-being: there was a tacit acceptance that no negative movement in these areas would be acceptable, even if authenticity and validity was improved.
My overall observation of the papers presented at the conference was that individual lecturers, faculties and universities are searching for ways to make changes to their assessment designs in the search for new ‘traditional’ methods - in other words, assessment approaches that are sufficiently authentic to maintain stakeholder confidence, but can be used en-masse across large departments and faculties.
Ironically, other innovations in assessment using AI - like auto-marking or task design may have workload implications that make increasingly authentic tasks more plausible. If AI is able to take on some of the load of making judgements on produced work from students, then lecturers might be able to spend that time operationalising interactive tasks like vivas and collaborative work.
As Higher Education and other institutions explore ways to innovate, it's important to mitigate risks of change that aren't grounded in sound assessment understanding and theory. As educational leaders are tasked with disseminating best practice to colleagues, it is crucial for those leaders to remember that for colleagues to be truly empowered - and have confidence in the evolution of their assessments - they must have a bedrock of powerful assessment knowledge to draw upon, which then allows them to justify their decisions."
The Assessment Network is part of Cambridge University Press & Assessment. We provide professional development for impactful assessment.