Is a university degree still worth the paper it is written on? That rather depends on whether a student is graded on their own work or on that of an AI passed off as their own. With three-quarters of universities continuing to make use of remote, uninvigilated ‘exams’, that question answers itself.
The hundreds of thousands of students who are receiving their A-Level and GCSE results this month do so with the confidence that their grades reflect a gold standard in assessment. With standardised national assessments, produced by national exam boards, regulated by Ofqual and administered primarily via invigilated, in-person exams, our system of school assessments is designed to maximise fairness, comparability and integrity.
In contrast, each university is responsible for determining not merely its own curriculum and assessment methods but its own grade boundaries, with minimal oversight from the light-touch external examiner system and regulator. It is hardly surprising that grade inflation is rife, with the proportion of first-class degrees awarded rising from 13 per cent to 30 per cent in under two decades – despite a fall in the ability of the average undergraduate. But with the rise of generative AI, what was once a chronic problem has become acute.
This invidious system puts students in a no-win situation
A new report by Professor Philip Newton, published by Policy Exchange, reveals that around three in four UK universities are using online, remote exams for assessments that contribute to a student’s final grade. 70 per cent plan to continue with this type of assessment – while only 10 per cent use online invigilation for all their online remote exams. Typically introduced during the Covid pandemic, many universities – including some of the most distinguished – have persisted with these flawed assessment methods, with the only guard against cheating being to ask students to refrain from doing so.
Of course, it was always possible for a student to cheat in such exams by taking them with the aid of another person. But generative AI has dramatically lowered the barriers to dishonesty by placing a highly able ‘assistant’ at the fingertips of every student. Coursework is also vulnerable to cheating, yet over 90 per cent of students say they have been assessed by means of a take-home essay.
This invidious system puts students in a no-win situation. Generative AI use is strongly incentivised, its use effectively connived at. It is trivial to access and doing so will enable any student to improve their mark. Those students who do follow the ‘honour codes’ set out by universities will therefore be penalised with lower grades and a longer-term disadvantage as they enter the labour market – where they may, with dark irony, have their job applications rejected on the basis of those lower grades by recruiters who have themselves outsourced the screening process to AI.
Staff are also in a no-win situation. They know students are using AI: in a recent survey, 94 per cent of students reported using AI in assessments, though not all of these will have been cheating. Yet there is no requirement for them to be trained in effective assessment mechanisms, nor will they necessarily be rewarded for undertaking the difficult, nigh-on-impossible task of detecting which students have used AI in a system where university administrators prioritise student satisfaction, league table position and attracting lucrative international students.
For an academic system whose credibility is already reeling from multiple blows, the looming AI assessment scandal reinforces the hypothesis that, for too many universities, maintaining academic standards is simply not a priority. The complacency with which they have chosen to retain obviously substandard assessment practices in the age of AI is flabbergasting. Recent events, such as the prestigious Jardine Foundation’s decision to cease future scholarships for students to study in the UK in favour of leading universities in Asia, demonstrate that UK universities can no longer take their world class status for granted.
Domestically, if employers cannot be confident that degrees mean what they claim to, they will increasingly bypass them. Already, some leading graduate employers have dropped their degree requirement in recruitment, relying instead upon their own assessments. And in 2021, the Solicitors Regulation Authority launched the Solicitors Qualifying Examination, saying:
We cannot know from the current system of legal education that all aspiring solicitors are assessed to a consistent standard and achieve the same outcomes.
As Sir Vince Cable, former secretary of state for business, innovation and skills, wrote in a foreword to the Policy Exchange report:
Fundamental to the valued place of universities in our society is their role in maintaining confidence in the status and value of a degree, as a demonstration of the learning that has taken place. This confidence rests on integrity but also on third-party validation. This is especially so when there are repeated reports of degree grade inflation and in the aftermath of the recent big expansion in numbers of degree-awarding institutions and students.
The report sets out a number of proposals to safeguard academic standards in assessment, including requiring the immediate cessation of online unsupervised remote exams – enforced, if needed, by the Office for Students – ensuring all assessment methods are AI-proof, and including more information about assessment methods on degree transcripts.
Unless action is taken, the AI assessment crisis could be the final straw for the UK’s higher education sector. Universities need to clean up their act – and fast – before it is too late.
