Taking Up Space: Can the Ubiquity of Artificial Intelligence Birth a Renaissance for Authentic Assessment?

Orlagh McCabe, Reader, Manchester Metropolitan University
Dr Eileen Pollard, Manchester Metropolitan University
Dr Stephen Powell, Manchester Metropolitan University

Authentic assessment isn’t new…

But in the aftermath of the impact of Generative AI platforms on Higher Education, authentic assessment is having a moment.

In our educational development roles, we have been thinking, talking and generally living-and-breathing the nail-biting anticipation of daily updates, while simultaneously attempting to reconcile the potential impact of AI on assessment practices at our institution, and across the sector more broadly. This blog very briefly summarises what we have concluded so far.

Authentic assessment of and for learning is conducted through ‘real world’ tasks requiring students to demonstrate their knowledge and skills in meaningful contexts (Swaffield, 2011). Through adopting findings from Verónica Villarroel’s systematic review (2018), colleagues can consider these three main dimensions to their authentic assessments in our changing world:

• Realism (having a realistic task and realistic context).
• Cognitive challenge.
• Feedback loops for evaluative judgement (developing students’ ability to judge quality).

Moreover, recent work from McArthur (2023) also urges us to pay closer attention to the social value of authentic assessment for broader applicability, which is enabled through a richer understanding of authenticity. To do this, McArthur suggests that we should further consider why the task matters and indicates that doing this will promote ‘a shift from the student in isolation to the student as a member of society’ (85).

However, colleagues may still choose to base their assessment design on real-world contexts or professional standards, both of which will involve AI for future graduates.

Whatever the chosen path, revisiting the fundamentals of constructive alignment, checking that the learning outcomes effectively describe what should be known or be able to be done sets a solid foundation for meaningful student experiences.

Finally, authentic assessment draws upon many of the features of active learning; therefore, our resources on active learning might also prove a useful starting point for assessment design.

Some examples of potentially AI-resistant assessment

As authentic assessment often draws on context, tasks or real-life situations, Generative AI is less likely to be able to produce specific content about it. Therefore, we have adapted Engstrom’s (2008) five general principles of autoethnographic writing, which Kahl (2013) recommended to assess reflective writing (cited in Seal et al, 2021).

They also now look potentially more AI-resilient for a reflective assignment today:

• Critically reflect upon prejudices that you, as a student, bring to the situation.
• Examine the effect that you have on the situation.
• Evaluate the role of ethics in your writing and interactions/learning.
• Discuss the impact your learning has on you.
• Reflect on what you have learned about power in society through your interactions/learning.

Colleagues may also find this link to our resource on assessment activities (created by Rachel Forsyth) useful for ideas about how to assess practice, use an interview or even an event to assess learning in our post-ChatGPT world.

AI, Authentic Assessment and Critical Pedagogy

Freire recognised that calls for objective, ‘neutral’ assessments (an effect more recently of neoliberalism) lead to teaching to the assessment, which he saw as recasting learning as ‘banking’, or learning-by-rote (1972:46). The destination for this approach to education is instrumentalising degrees to only be about getting a job. Or, as Giroux argued: ‘assessment prepares students for corporate life’ (2005:43).

The banking model requires assessments that simply reproduce the status quo, which are of course vulnerable to AI, as they are inherently generic and thus reproduceable. But now Generative AI literally reproduces the assignment too, without the student needing to think at all: the thinking is outsourced since all the information is ‘banked’ in the technology.

And this reading further explains how authentic assessments, such as autoethnographic approaches and dialogic interactions, hold the key to wrongfooting AI, since they interrupt this cycle of reproduction.

Therefore, as many have said, the disruption caused by Generative AI is actually an opportunity for us to craft fewer but more meaningful assessments of and for learning.

References

Friere, P. (1972) Pedagogy of the Oppressed. London: Penguin Books.

Giroux, H A (2005) The Terror of Neoliberalism: Rethinking the Significance of Cultural Politics. College Literature , 32(1): 1– 19.

McArthur, J., 2023. Rethinking authentic assessment: work, well-being, and society. Higher education, 85(1), pp.85-101.

Seal, M. et al. (2021) Enabling critical pedagogy in higher education. St Albans: Critical Publishing (Critical practice in higher education)

Swaffield, S. (2011) Getting to the heart of authentic Assessment for Learning, Assessment in Education: Principles, Policy & Practice, 18:4, 433-449, DOI: 10.1080/0969594X.2011.582838

Villarroel, V. et al (2018) Authentic assessment: creating a blueprint for course design, Assessment & Evaluation in Higher Education, 43:5, 840-854, DOI: 10.1080/02602938.2017.1412396