½¿É«µ¼º½

As artificial intelligence (AI) programs such as ChatGPT, Claude and Gemini become increasingly integrated into student work processes, universities are confronted with an uncomfortable question. If it is possible for generative models to generate well-written, cohesive essays in mere seconds, then what exactly are you grading when you mark an academic essay? The implications extend beyond plagiarism detection. They challenge us to reconsider what it is that constitutes learning, authorship and the very nature of assessment in higher education.

‘If it is possible for generative models to generate well-written, cohesive essays in mere seconds, then what exactly are you grading when you mark an academic essay?’

Academic essay writing has been the mainstay of higher education evaluation for some time now because of its capability to show critical thinking, argumentation, synthesis and independent research skills. It is not merely a genre but an approach to learning. Authors like Lea and Street (1998) have demonstrated that writing essays is an integral feature of the epistemological activities of the university. It is intended to teach learners to make claims for knowledge and respond to sources in conversation, as well as build disciplinary identities.

But AI complicates this model in fundamental ways. If students can now generate well-polished, grammatically correct essays without any complex critical thinking or effort involved, then the essay as an indicator of that thinking begins to erode. Something that was previously assumed to be an exercise in intellectual work is now an exercise in prompt engineering ability or editorial skill. What, precisely, are we assessing when we ask students to submit written work? Are we evaluating their grasp of content, their ability to write in disciplinary genres, or their capacity to think critically and independently? It is not merely a crisis of authenticity; it is one of definition. If learning is the acquisition of knowledge to be reasoned with and applied, and the student can describe those things discursively or in debate, does it make a difference that the student’s paper was written in collaboration with AI? Are we measuring knowledge with the essay, or are we measuring the skill to execute a specific academic genre in specific form?

Generative AI challenges the field of education to reconsider the tension between learning outcomes and the practices of assessment. According to Eaton (2023), academic integrity is not to be regarded as an absolute rule but as an institutionalised practice based on culture and pedagogy. This strategy reframes the ethical challenges of AI not as issues to solve through detection and punishment but as possibilities to redesign learning spaces that promote openness, reflection and responsibility. From this perspective, the essential concern is not to stop the use of AI but to develop assessment processes that welcome integrity through substantive engagement and transparency.

Instead of falling back on surveillance or reverting to classroom exams, universities can use the current moment to expand their repertoires of assessment. This does not involve the discarding of the written word; it is merely reinventing its function in the process of learning. Written assessment can continue to be useful; however, it might need to be supplemented by reflective commentaries, oral defences or electronic portfolios that show the process along with the product.

Incorporating the use of AI in the curriculum also presents a more productive option. When learners have an invitation to use AI tools in an open manner, annotating its use, justifying its selection and thinking through the tool’s limitations, they are able to build metacognition and digital literacy skills. This way, AI is not a short cut but a site of critical reflection. Alternative forms of assessments such as group work, podcasts, policy briefs or graphical descriptions can capture other sides of learning as well. These modes will be less susceptible to AI co-authorship and will be optimised for those multimodal, interdisciplinary ideals of the modern world. Most importantly, these changes need to be paired with inclusive pedagogy and explicit scaffolding to engage and not hinder learners.

‘Crossing beyond the page does not necessarily involve abandoning it, but it does involve the acknowledgment that learning can be expressed, recorded and established in more diverse, dialogical and dynamic modes of expression.’

Higher education stands at a critical juncture. AI is not an outside force to be kept at bay but an evolving state of knowledge production that has the ability to be assimilated with reflection. The challenge is not one of whether essays will survive but of whether teaching practices are adaptable, fair and resilient enough to respond to the challenges of an evolving epistemic environment. University assessment has always mirrored deeper beliefs regarding what is considered learning and whose learning is to be demonstrated. In the AI era, the challenge is to test those beliefs with greater urgency. Crossing beyond the page does not necessarily involve abandoning it, but it does involve the acknowledgment that learning can be expressed, recorded and established in more diverse, dialogical and dynamic modes of expression.

References

Eaton, S. E. (2023). Postplagiarism: Transdisciplinary ethics and integrity in the age of artificial intelligence and neurotechnology. International Journal for Educational Integrity, 19, 23.

Lea, M. R., & Street, B. V. (1998). Student writing in higher education: An academic literacies approach. Studies in Higher Education, 23(2), 157–172.