At the click of a mouse button the essay writing begins. It is happening much faster than I can read it. By the time the author is penning the conclusion, I have barely digested the opening sentences of the introduction.
‘If you’d like any sections expanded or refined, let me know!’ adds the writer at the bottom of the thesis.
Examining its contents proves both sobering and galling. As a sixth-year pupil in the 1980s, it took me weeks to produce a dissertation on the subject at hand here.
Before putting pen to paper, I had to read six plays and familiarise myself with a wealth of literary criticism on all of them.
Now a competent study, comparing and contrasting the use of menace as a dramatic device in the plays of Harold Pinter and Joe Orton, has been produced in seconds by an artificial intelligence (AI) program.
It finds apposite quotes from the work of both dramatists to support its points. It correctly identifies the distinct strains of menace in which each playwright specialised, noting that Pinter’s was psychological while Orton’s was rooted in casual brutality.
There is just one problem. The essay has a section on the Pinter play The Homecoming, which was not on my reading list, and says nothing about The Dumb Waiter, which was.
It’s feared the rise of AI could have a detrimental impact on education
I inform the writer and, seconds later, it produces a revised draft, deleting reference to The Homecoming and inserting a credible reflection on menace in The Dumb Waiter.
I experiment further, ordering an essay analysing the varying moods in the early verse of French poet Arthur Rimbaud – the theme of my senior honours dissertation. Again, it arrives within seconds and is disconcertingly competent.
For those whose student years are long behind them, testing AI’s essay-writing ability can prove an education. But for our current crop of students, could AI be allowing them to bypass education almost entirely and still gain degrees?
Days ago, a freedom of information inquiry revealed a 700 per cent increase in academic misconduct cases at Scottish universities involving the use of AI to produce course work.
The politician who obtained the data, Scottish Conservative education spokesman Miles Briggs, believes that the 1,051 cases of AI ‘cheating’ during 2023-24 are ‘the tip of the iceberg’.
And, given that a report last month found 92 per cent of students are using AI as a study aid, few in academia are minded to challenge his claim.
Indeed, many acknowledge that the implications for both secondary schools and further education are seismic. Simply put, if a tool can do students’ course work for them, how can they be assessed?
And how truly indicative of their academic abilities are the degrees they leave with?
Scottish Conservative education spokesman Miles Briggs, believes that the 1,051 cases of AI ‘cheating’ during 2023-24 are ‘the tip of the iceberg’
The biggest worry, perhaps, is AI is improving the standard of its work far more rapidly than universities are finding ways to defend the integrity of their courses. As one academic puts it, AI’s efforts ‘are starting to cross into pass territory’. It is getting cleverer, its writing noticeably crisper than even a year ago.
‘We’re probably at the point at which the written essay is no longer a viable manner of assessment,’ concludes Andres Guadamuz, an intellectual property law expert at the University of Sussex.
The methods by which students can lighten their academic load are alarmingly simple. First they install an app such as ChatGPT on their phones – or even just go to its website. ‘What can I help with?’ it says on its welcome screen. I typed in: ‘Write an academic essay on the theme of menace in the plays of Harold Pinter and Joe Orton. Discuss how the playwrights use menace to create drama.’
That question was simply handled. The tougher question schools and universities must grapple with is this: what would stop any student passing any essay task on any subject over to AI?
For now, perhaps, the biggest disincentive may be the fear of discovery.
Academics with experience of marking AI-generated work passed off as a student’s own point out the technology’s tendency to ‘hallucinate’. It can invent quotes – even make up entire texts and, most unhelpfully, insert them in bibliographies.
For a clear reading on its competence, suggests Andrew Moran, professor of politics and international relations at London Metropolitan University, ask it to expound on a subject or person you know a lot about and count the errors.
‘It will often create things that are simply not true, or it will match information from different people,’ he says.
University of Edinburgh, Professor Lindsay Paterson believes the widespread use of AI by students makes continuous assessment ‘wholly unreliable’
The problem is AI is steadily correcting its own mistakes. How long before it is practically infallible?
The supreme irony is that Turnitin, the AI detection software used by Scottish universities to unmask the cheats, uses elements of AI and machine learning techniques to do so.
Thus, in an absurdist technological arms race, AI is deployed to police itself.
At Abertay University – which dealt with a record 351 cases of ‘unacceptable AI use’ last year – sociology and criminology lecturer Dr Stuart Waiton reports that, in his department, AI wrote an ‘excellent’ academic project stretching to 30 pages only this week.
He says: ‘AI could be a brilliant tool for postgraduates and researchers. If you already have a level of expertise and knowledge, AI can be used to enhance research.
‘The problem is that if it is used by undergraduates, it becomes a way of avoiding having to become an expert in the first place – a cheat – that means you never have to read a book or think.
‘A key potential problem with universities is whether or not they take AI cheating seriously or simply ignore the cheating dimension of using it.’
Dr Waiton adds: ‘Personally, I think any use of AI should be classed as plagiarism and the essay failed – I suspect no institution relates to it in this way.’
The issue is exacerbated by the fact that, in recent decades, universities have moved towards a continuous assessment model, placing less emphasis on end of term exams. This only enhances the scope for AI use.
Dr Waiton points out that some institutions are even ditching ‘in-person’ exams in favour of online ones.
He says: ‘This means that students sat in their bedrooms can use AI even here, and the idea of a degree potentially becomes a mockery.
‘Of course, having proper exams is the obvious solution to this and again, personally, I would be happy to see degrees awarded based largely on exams, but universities, nervous about the “student experience” or, indeed, putting real pressure onto students, are unlikely to adopt this even though it would largely solve the AI issue.
‘And that is really the issue. Do our universities still believe in standards and are they
prepared to do what is necessary to make Scottish universities the best, potentially, in the world? I’m not holding my breath.’
At the University of Edinburgh, Professor Lindsay Paterson believes the widespread use of AI by students makes continuous assessment ‘wholly unreliable’.
He says: ‘There is literally no way of telling whether a piece of written work was genuinely produced by the student or was generated by AI.’
He sees only two possible solutions. The first is to add an oral examination to every piece of written work – a massively labour-intensive exercise which would result in weeks of extra work for university staff.
Professor Paterson adds: ‘If oral assessment is mostly not feasible, then the only alternative that could safeguard standards would be to return to using invigilated examinations, which most university courses have dropped in recent years.
‘I think this would be a good idea for all sorts of reasons, not just dealing with AI.’
If there is an upside to AI-generated essays, it may be that they signal the death knell for the unscrupulous ‘essay mill’ industry which encouraged students to pay to have their course work done for them by anonymous academics, often based overseas.
But Professor Paterson says universities’ ‘lack of urgency’ in dealing with that problem ‘does not encourage optimism’ that they will deal effectively with AI.
He adds: ‘If they won’t, then governments at Holyrood and Westminster must, for the sake of the international reputation not only of UK universities but also of the UK as a society.’
Is the Scottish Government, then, on top of the issue? According to MSP Mr Briggs, who uncovered the latest shocking data: ‘I don’t think they really know what to do, is the truth of the matter.
‘They haven’t done the deep dive into what is happening in schools and universities on this and what it means to the grading system.’
The reality, he suggests, is that the ‘genie is already out of the bottle’. He adds: ‘There is no point in us just clutching our pearls on this. We’ve got to look towards how are universities going to move with the times and accept this is happening – and start to assess what level young people are at in a different way.’ Academics such as Professor Moran now fear a certain inevitability in their students using AI.
In practice, he suggests, most will not submit the AI essay word for word but edit and expand it.
Yet he says the process is in danger of turning knowledge into a ‘convenience food’.
‘It is available at the touch of a button online, is effectively delivered to your door and there are so many outlets to choose from.’
Professor Moran adds: ‘We are rapidly losing ground to profound societal changes that could have unimaginable consequences for universities if we do not respond quickly.’
The consequences are, of course, being felt well beyond the realm of education.
Britain’s creative industries are under threat from AI as tech giants hoover up artists’ online content – songs, writing, pictures, the lot – to improve their AI models.
And, under Labour government plans, they will be able to do so while ignoring the copyright laws which ensure creators get paid.
Is this ill-gotten content now to be deployed by students to con their way to degrees they do not deserve?
Opinions differ on the ‘acceptable’ ways for students to use AI in their studies.
Feedback from undergraduates in last month’s Student Generative AI Survey 2025 finds that, besides the 18 per cent who admitted using it in submitted work, others use it to explain concepts, suggest research ideas, make grammatical improvements and crunch down lengthy articles into a summary of key points.
But, asks, Professor Moran: ‘How can you know that AI has generated the correct information if you’ve not actually read it yourself?’
He makes a point of telling his first-year students to walk across to the library and ‘breathe in the books’.
‘I say to them people have spent years researching the information in those books for you, for you to use, and you must do that.’
He fears that, for too many students, knowledge is a ‘thing on your phone’ rather than something to be sought out in the world of books.
And how are former students supposed to feel? The ones who sweated for weeks or months over dissertations, starting with a blank sheet of paper, when AI can now write reams of intelligent insight in seconds?
Well, suggests Professor Paterson, we can console ourselves with the knowledge that we engaged in true learning.
He says: ‘The process of discovering information, and then collating, analysing and writing about it is the truly educative task.
‘That’s why we remember these things so well – not necessarily what we found, but the mental and emotional processes by which we found it. That is a character-creating experience.
‘Students who think they can skip this by using AI are deluding themselves. I hope that employers realise that too, and make some effort to find out whether a student’s apparently brilliant grade was due to their own work.’
Universities Scotland, the umbrella body for the nation’s higher education, says universities are ‘vigilant’ regarding the use of AI and will continue to evolve methods of assessment to uphold academic integrity.
A spokesman adds: ‘Whilst the use of AI does bring challenges in certain contexts, it also brings some benefits to the learning environment in a university.
‘Not least the fact that employers increasingly expect graduates to have AI skills, as its use also becomes more prevalent in the workplace.’
A Scottish government spokesman says: ‘The Scottish Funding Council and Quality Assurance Agency are continuing to work with universities to address the risks and opportunities presented by artificial intelligence software, and its potential impact on assessments, academic integrity and standards.’
Naturally, ChatGPT has a line on this too. Asked whether the use of AI threatens the integrity of students’ degrees, the answer it generated allowed that it can do ‘if it leads to over-reliance on automated tools, bypassing critical thinking and personal effort’.
This, it said, ‘undermines the learning process and may devalue the degree’.
Clear guidelines should be encouraged, it stated, to ensure that AI ‘complements rather than replaces student effort’.
It seems as if the software may not entirely approve of itself. However, its invitation to students – and, indeed, everybody – remains an open one.
‘Ask anything,’ it urges.
