21/09/2026
AI in University: So What Are We Assessing, the Student or the Technology?
We do not yet know what university education will look like in a world shaped by AI, but in the meantime, how do we know what our students can do? AI is becoming increasingly difficult to control in the English language classroom. Assignments, research projects and presentations are increasingly being produced or polished with generative AI. Students are also becoming increasingly savvy, asking AI to include mistakes, simplify language or write at their expected level.
AI is no longer a marginal issue in higher education. It is now close to universal. The latest HEPI Student Generative AI Survey found that 95% of students use AI in some form, while 94% use generative AI to help with assessed work.
For international students, the issue can go even further. Jisc reports that non English speaking students are using tools such as DeepL and Google Translate to translate texts and support their understanding. Students are also using AI to generate ideas, structure arguments, proofread and improve their written work.
This creates a fundamental problem for language assessment: how do we know whether we are assessing the student’s English or the technology they have used?
At Academic English UK, we have started developing handwritten, timed assessments using unseen academic sourcesas a practical response. Without access to AI, students have to read, think, organise and write for themselves. These are key skills in academic language development and are reflected in the CEFR, including the ability to process information, synthesise ideas and communicate meaning effectively.
We do not yet know what university education and assessment will look like in five or ten years. AI is developing rapidly, and universities are still working out how it should be used, taught and assessed. But while that future is still uncertain, we need reliable ways to understand what our students can actually do.
For EAP and second language learners, a simple paper and pen assessment using unseen material provides one way of doing this. It gives teachers an opportunity to see the vocabulary, grammar, sentence control, cohesion and academic expression students can produce independently.
https://academic-englishuk.com/academic-source-based-writing-tests/
The future of education with AI remains uncertain. For now, perhaps the best approach is to embrace the possibilities of AI while also making sure we do not lose sight of the value of seeing students think and write for themselves. Sometimes, that means putting the technology aside, giving them a sheet of paper and seeing what they can do.
Chris Wilson
AEUK
Sources:
https://www.hepi.ac.uk/reports/student-generative-ai-survey-2026/
https://www.jisc.ac.uk/reports/student-perceptions-of-ai-2025