AI has exposed weaknesses in university assessment. Turning back the clock is not the answer

Policy Exchange is right to draw attention to the need for universities to ensure that degrees provide trustworthy evidence of what graduates know and can do in the recent report, Evidence of Learning Through Assessment: Protecting the Value of a Degree in the Age of AI. Remote exams without effective safeguards and take-home essays can be vulnerable to inappropriate AI use. Simply telling students not to use AI is not a credible strategy where those rules cannot be monitored or enforced.

We also support the report’s calls for better assessment training, greater transparency and a wider range of assessment methods. Students, employers and the public should be able to understand what a graduate has demonstrated and the evidence on which that judgement is based. Every institution should also have clear guidance on when AI may be used, how it should be acknowledged and what constitutes misconduct.

The e-Assessment Association, however, fundamentally disagrees that the answer is to make face-to-face assessment the norm or return predominantly to in-person examinations.

The report argues that in-person examinations should replace remote assessment unless remote assessment is fully secure. Our response is clear: where remote assessment is not sufficiently secure, we should make it secure, not abandon it.

AI is here to stay. It is already changing education, employment and society. Universities cannot respond by designing assessment as though AI does not exist or by recreating a system developed for a pre-digital world.

Patrick Coates, CEO of the e-Assessment Association, said: “Similar concerns were raised when calculators became widely available and again when students began using personal computers. The answer was not to exclude those technologies or return to earlier ways of working, but to rethink what students needed to learn and how that learning should be assessed.
“AI presents a far greater challenge, but the principle remains the same: we must adapt assessment so that it reflects the world in which students will live and work, while continuing to provide trustworthy evidence of their own knowledge, skills and judgement.”

Students must still develop independent knowledge, critical thinking and professional judgement. Universities must be able to demonstrate that these capabilities belong to the student rather than to the technology they have used. But an exam hall is not the only place where that evidence can be gathered.

Nor is an in-person examination automatically valid, fair or secure. It may confirm who is sitting the examination, but it can still test memory rather than understanding and provide limited evidence of whether a student can analyse, communicate, solve problems or apply knowledge.

Remote assessment can be secured through an appropriate combination of identity verification, online invigilation, secure platforms, controlled access to resources, effective assessment design and human review. These measures must consider privacy, accessibility, fairness and the student experience, but the technology and expertise already exist.

The central question should not be whether an assessment is online or in person, but whether it provides sufficient, trustworthy evidence that an individual has achieved the intended learning outcomes.

This will normally require a combination of methods. Digital assessment offers significant advantages over pen-and-paper exams, including more accessible and authentic tasks, interactive and multimedia content, faster feedback and richer evidence of how students apply their knowledge and skills. Alongside this, practical demonstrations, vivas, presentations, simulations and portfolios can provide evidence of capabilities that traditional examinations may not capture.

A modern assessment system must also assess students’ ability to use AI critically, responsibly and transparently. Graduates should be able to demonstrate both what they can do independently and what they can achieve using the tools that will form part of their future lives and careers.

The report’s foreword calls for “AI-proof” assessment, while its recommendations refer to methods that are demonstrably resistant to generative AI. Assessment should certainly be resilient to inappropriate AI use, but attempting simply to exclude AI is neither realistic nor sufficient.

Generative AI has exposed genuine weaknesses in university assessment. Those weaknesses must be addressed, but the answer is not to retreat from digital assessment or return to the assessment system of the past.

We should not respond to the defining technology of our time by returning to the assessment system of the past. We should invest in secure, modern and evidence-led assessment that demonstrates what students know, what they can do independently and how responsibly they can use AI.

About the eAA

The e-Assessment Association (eAA) is the global professional membership body dedicated to advancing the use of technology in assessment. Bringing together awarding organisations, educational institutions, professional bodies, employers, technology providers, researchers and independent experts, the eAA provides a trusted forum for collaboration, knowledge sharing and innovation across education, professional certification and workplace assessment.

The Association supports the development of high-quality, fair and accessible digital assessment through research, best practice, Special Interest Groups, international conferences, webinars and the annual International e-Assessment Awards. By connecting expertise from across the sector, the eAA helps shape the future of assessment and promotes the responsible adoption of technologies, including artificial intelligence, to improve outcomes for learners and assessment providers worldwide.

Individuals can join the eAA for free. Find out more here: www.e-assessment.com/membership