Tag: case

I loved magic as a kid. Card tricks, disappearing coins, little felt rabbits in pretend top hats. “Now anyone can be a magician,” proclaimed the advert in the Argos catalogue. Ta da. Now that’s magic.

I’d make pretend tickets, rearrange the seating in the front room, and perform shows for the family – slowly learning the dark arts of misdirection and manipulation along the way. When I performed, I generated pride.

Over time I found that some of those skills could be used to influence people more generally – to make them feel better about themselves, to change their decisions, to trigger some kind of behaviour.

Sometimes, I’d rationalise, as long as I was doing it for the right reasons, it was better if they didn’t know it was a trick. The end justified the means. Or did it?

People love it when they know that magic is being performed as magic – the willing suspension of disbelief, the pleasure of being fooled by someone who’s earned the right to fool us. When they give permission to be illegitimately impressed, all is fine.

But what they can’t stand is being lied to. We don’t like being deceived. Most political news in this country centres on who lied and about what. We’re obsessed with it.

The cover-up is always worse than the crime, yet everyone still does it – they have to, they rationalise, to keep up, or to get permission. The gap between how things are and how we present them is the game.

Once they’re in, it won’t matter that the sector painted an unobtainable picture of student life for applicants. Once the funding is secured, universities can fess up that it isn’t as good as government thought it would be later. Once the rules are published, better to ask for forgiveness over the impact on net migration – not permission.

I think about that little magic set I got a lot, because so much of what AI does still sits for me in that “magic trick” space.

Ta da. Look what it can do. Generate an essay, write a play, create some code, produce an image of the Pope in a puffer jacket. But the line between magic and lies is a slippery slope, because its number one use case is pretence.

AI is used to lie – fake essays, fake expertise, fake competence. But mostly to make us look better, appear faster and seem wiser. The anxiety about being “found out” is the anxiety of the liar, not the audience at a magic show. Students worry they’ll be caught. Universities worry their degrees will be worthless.

Everyone worries that the whole edifice of qualifications and signals and “I know something you don’t” will collapse under the weight of its own pretending. But the pretending was already there – AI just makes the tricks cheaper, and much harder to sustain.

When I look back upon my life

I’ve been in a particularly reflective mood recently – I turned 50 at the weekend (I can’t believe it either, it’s the moisturiser) and there’s something I can’t shake. When I look back upon my life, it’s always with a sense of shame.

When I got accepted to the University of the West of England in the mid-nineties, grandparents on both sides were thrilled that I had “got into Bristol”. A few extra Bonusprint copies of the sunken lawn at the St Matthias campus helped.

It hadn’t started as a deliberate lie – more a misunderstanding about where we had driven to on open days – but instead of correcting it, I doubled down.

Nobody in my family had been to university, and I doubt they would have discerned the difference. But on some level I thought I had to prove that their financial support was for something rare. Something… special.

Decades later I realised that the entire edifice of higher education runs on the same kind of slippage – the gap between what universities actually do, and the status they are assumed to have and confer.

Applicants and their families celebrate “getting in” as if admission itself were the achievement. Parents frame graduation photos, the ceremony mattering more than the three years that preceded it. Employers use degree classifications as sorting mechanisms while moaning that the sort has not delivered the graduates they wanted. There’s a graduate premium. And so on.

Those of us who write about higher education are no better. Our business model rests on “I know something you do not” – the insider knowledge, the things you haven’t noticed, the analysis you can’t get elsewhere. Scarcity of information, monetised. I’ve built a career on being the person in the room who has read the regulatory guidance.

But now, suddenly, a machine can summarise the guidance in seconds. Not as well as I can – not yet, not always – but well enough to make me wonder what I am actually for. What value I bring. How good I am at… pretending.

AI doesn’t create that anxiety. It exposes something that was always there – the fear that our value was never in what we knew, but in other people not knowing it. And that eventually, someone might find that out.

It’s always with a sense of shame

Back in 1995 my first (handwritten) university essay was about the way the internet lets you become someone you are not. Chatrooms were new and identity was suddenly fluid. You could lie about everything – your age, your appearance, your expertise – and checking was hard.

The internet has been flooded with exaggeration ever since. Wish.com tat that looks nothing like the picture. LinkedIn profiles that bear no relationship to actual jobs. Influencers selling lives they don’t live in places that barely exist.

But it has also liberated us. At UWE, I lived through the transition from index cards in libraries to DogPile, asking Jeeves and Google. The skill of navigating a card catalogue, of knowing which reference books to check – it felt essential, and then it was worthless. For one semester, we were told we weren’t allowed to use search engines. The faculty held on for a while, then let go.

In my first year, I chose a module involving audio editing on reel-to-reel tape. Splicing, cutting, winding, knives. At the end of the year, I got a job helping to put the equipment in a skip. The skills I’d learned were obsolete before I graduated.

Each time, there was a period of pretending that the old skills still mattered. Each time, the system eventually admitted they didn’t. Each time, something was revealed about what had actually been valuable all along. The card catalogue wasn’t the point – finding and evaluating information was. The handwriting wasn’t the point – thinking under pressure was. The reel-to-reel wasn’t the point – understanding how to shape a story with sound was.

Now the sector clings on to exams, essays, and the whole apparatus of assessment that assumes that producing a thing proves you learned something. The system holds on – but for what?

I’ve always been the one to blame

If I rummage through the AI pitches that land at [email protected], I can see a familiar pattern.

There are catastrophists. Students are cheating on an industrial scale. The essay is dead. Standards are collapsing and students are cognitively offloading while the great plagiarism machine works its magic.

There are tech evangelists. Productivity gains, personalised learning, democratised access and emancipation – just so long as you don’t ask who is selling the tools, who is buying the data, or what happens to students who can’t afford the premium tier.

Then there is the centrist-Dad middle. “It is neither all good nor all bad” – balance, nuance, thoughtful engagement, and very little about what any of this is actually for.

The catastrophists are wrong because they assume what’s being bypassed was valuable – that the essay-writing, the exam-sitting, the problem-set-completing were the point rather than proxies for something else. If the activities can be replaced by a machine, what were they measuring?

The evangelists are wrong because they assume more efficiency is always better – that if AI frees us from X, we’ll have more time to do Y. But they never say what Y is. Or whose time it becomes. In practice, we know – the efficiency dividend flows upward, and never shows up as an afternoon off.

The balanced view is just as bad, because it pretends there’s no choice to be made. It lets us sound reasonable while avoiding the harder question – what is higher education for?

At the high risk of becoming one of those bores at a conference whose “question” is a speech about that very issue, I do think there is a choice to be made. We ought at least to ask if universities exist to sort and qualify, or to form and transform. AI forces the question.

For everything I long to do

Let’s first admit a secret that would get me thrown out of the Magic Circle. The industrial model of education was built on scarcity, and scarcity made a certain kind of pretending possible.

Information was scarce – held in libraries, transmitted by experts, accessible only to those who got through the door. A degree meant three years in proximity to information others could not reach.

Attention was scarce – one lecturer, two hundred students, maybe a weekly seminar. The economics of mass higher education turned teaching into broadcast, not dialogue, but the scarcity, coupled with outcomes stats from the past, still conferred value.

Feedback was scarce – assignments returned weeks later with a grade and a short paragraph. The delay and brevity made the judgement feel weighty, even oracular.

In a scarcity system, hoarding makes sense. Knowledge is power precisely because others don’t have it. “I know something you do not” isn’t a bug – it’s the business model. But once something isn’t scarce any more, we have to search again for value.

We’ve been here before. Calculators didn’t destroy maths – they revealed that arithmetic wasn’t the point. Google didn’t destroy research – it revealed that finding information wasn’t really the hard bit. Each time the anxiety was the same – students will cheat, standards will collapse, the thing we valued will be lost. Each time the pretending got harder to sustain.

For me AI fits the pattern. Not because it knows everything – it obviously doesn’t. Its confident wrongness is one of its most dangerous features. But it makes a certain kind of information effectively free. Facts, frameworks, standard analyses are now available to anyone with an internet connection and the wit to ask.

And it hurts to carefully build and defend systems that confer status on things humans can do – only to have something come along and relieve humans from having to do them. It causes a confrontation – with value.

No matter when or where or who

During the early days of Covid, I came across a Harvard Business School theory called Jobs To Be Done. People pay to get a job done, but organisations often misunderstand the real job they’re being paid to perform.

As a kid, the Sinclair ZX Spectrum in our house was marketed as an educational tool – an invitation to become a programmer. Some did. Most, like me, worked out how to make the screen say rude words and then played games.

Students have at least two jobs they want done. One is access to well-paid and meaningful work, made possible through obtaining a degree and supplied by academic programmes. The second is coming of age – the intoxicating combination of growing up and lifestyle. Becoming someone. Finding your people. Working out who you are when you’re not defined by your parents or your school.

Universities have always provided both, but only dare attribute value to the first. The second is treated as incidental – “the student experience”, something that happens around the edges. But for many students, perhaps most, the second job is why they came. The qualification is the price of admission to three years of transformation.

AI increasingly handles the first job – the information, the credentials, the sorting – more efficiently than universities ever could. If that were all universities offered, they’d already be obsolete. What AI can’t provide is the second job. It can’t help us become someone. It can’t introduce us to people who will change our lives. It can’t hold us accountable, or surprise us, or make us brave.

During Covid, I argued that universities should cancel as much face-to-face teaching as possible – because it wasn’t working anyway – but keep campuses open. Not for teaching – for being. For studying together, bonding, bridging, belonging.

I’ve not changed my view. AI just makes it more urgent. If the content delivery can be automated, the campus has to be for something else. That something else is formation.

Has one thing in common, too

A couple of years ago I came across Thomas Basbøll, resident writing consultant at Copenhagen Business School Library. He argues that when a human performs a cognitively sophisticated task – writes a compelling essay, analyses a complex case, synthesises disparate sources – we infer underlying competence. The performance becomes evidence of something deeper.

When a machine performs the same task, we can’t make the inference. The machine has processes that produce outputs. It doesn’t “know” anything – it predicts tokens. The output might resemble what a knowledgeable human would produce, but it proves nothing about understanding.

Education has always used performance as a proxy for competence. Higher education sets essays because it assumed that producing a good one required learning something. There was trust in the inference from output to understanding, and AI breaks it. The performance proves nothing.

For many students, the performance was already disconnected from competence. Dave Cormier, from the University of Prince Edward Island, described the experience of essay writing in the search era as:

have an argument, do a search for a quote that supports that position, pop the paper into Zotero to get the citation right, pop it in the paper. No reading for context. No real idea what the paper was even about.”

There was always pretending. AI just automated it.

Basbøll’s question still haunts me. What is it that we want students to be able to do on their own? Not “should we allow ChatGPT” – that battle is lost. What capacities, developed through practice and evidenced in assessment, do we actually care about?

If the answer is that appearing literate is enough, then we might as well hand the whole thing to the machines. If the answer is that we want students to actually develop capacities, then universities will need to watch students doing things – synchronous engagement, supervised practice, assessment that can’t be outsourced. A shift that feels too resource-intensive for the funding model.

What’s missing from both options is that neither is really about learning. One is about performing competence, the other is about proving competence under surveillance, but both still treat the output as the point. The system can’t ask what students actually learned, because it was never designed to find out. It was designed to sort.

Everything I’ve ever done

How hard should education be? The “meritocracy of difficulty” ties academic value to how hard a course is to survive – dense content, heavy workloads, high-stakes assessment used to filter and sort rather than support students. Go too far in the other direction, and it’s a pointless prizes-for-all game in which nobody learns a thing.

Maybe the sorting and the signalling is the problem. The degree classification system was designed for an elite era where classification signalled that the graduate was better than other people. First class – exceptional. Third – joker. The whole apparatus assumes that the point of education is to prove that your Dad’s better than my Dad. See also the TEF.

Everyone pretends about the workload. The credit system assumes thirty-five to forty hours per week for a full-time student. Students aren’t studying for anything like that. The gap is vast, everyone knows it, and nobody says it out loud because saying it would expose the fiction.

AI intensifies it all. If students can automate the drudgery, they will – not because they’re lazy, but because they’re rational actors in a system that rewards outputs over process. If the system says “produce this essay” and the essay can be produced in ten minutes, why would anyone spend ten hours?

Mark Twain might have said that he would never let his schooling interfere with his education. Today’s undergraduates would more often lament that they don’t can’t their lectures and seminars interfere with their part-time job that pays the rent.

Every place I’ve ever been

There’s a YouTube video about Czech railways that’s been stuck in my head for weeks now. They built a 200 km/h line between Prague and Budweis and held celebrations – the first domestic intercity service to break the 160 km/h barrier.

But only one train per day actually runs at that speed. It arrives late every time. Passengers spend the whole journey anxious about missing their ten-minute connection at the other end.

The Swiss do it differently. The Gotthard Base Tunnel was built for 230 km/h. Trains run at 200. The spare capacity isn’t wasted – it’s held in reserve. If a train enters the tunnel with a five-minute delay, it accelerates and emerges with only two. The tunnel eats delays. The result is connection punctuality of the kind where you almost always make your connection.

The Czech approach is speed fetishism – make the easily marketable number bigger, and assume that’s improvement. The Swiss approach is reliability – build in slack, prioritise the journey over the metric, make sure people get where they’re going.

It sometimes feels to me like UK universities have gone the Czech route. We’re the envy of the world on throughput – faster degrees, more students, tighter timetables, twelve-week modules with no room to fall behind.

But when anything goes wrong – and things always go wrong – students miss their connections. A bad week becomes a failed module. A failed module becomes a resit year. A mental health crisis becomes a dropout. Then we blame them for lacking resilience, as if the problem were their character rather than a system designed with no slack.

The formation model is the Swiss model. Slow down. Build in reserves. Let students recover from setbacks. Prioritise the journey over the metric. Accept that some things cannot be rushed.

At school they taught me how to be

Universities tell themselves similar lies about academics. It’s been obvious for a long time that the UK can’t sustain a system where researchers are also the teachers, the pastoral supporters, the markers and the administrators.

The all-rounder academic – brilliant at research, compelling in lectures, attentive in tutorials, wise in pastoral care, efficient at marking, engaged in knowledge exchange – was always a fantasy, tolerable only when student numbers were small enough to hide the gaps.

Massification stretched it. Every component became more complicated, with more onerous demands, while the mental model of what good looks like didn’t change. AI breaks it.

If students automate essay production, academics can automate feedback. We’re already seeing AI marking tools that claim to do in seconds what takes hours. If both sides are pretending – students pretending to write, academics pretending to read – what’s left?

The answer is – only the encounter. The tutorial where someone’s question makes you think again. The supervision where a half-formed idea gets taken seriously. The seminar where genuine disagreement produces genuine movement. The moments when people are present to each other, accountable to each other, and changed by each other.

They can’t be automated. They also can’t be scaled in the way the current model demands. You can’t have genuine encounters at a ratio of one to two hundred. Nor can you develop judgement in a twelve-week module delivered to students whose names you don’t know.

The alternative is differentiation – people who teach, people who research, people who coach, working in teams on longer-form problems rather than alone in offices marking scripts. But that requires admitting the all-rounder was always a lie, and restructuring everything around that admission.

So pure in thought and word and deed

If information is now abundant and feedback can be instant and personalised, then the scarcity model is dead. Good riddance. But abundance creates its own problems.

Without judgement, abundance is useless. Knowing that something is the case is increasingly cheap. Any idiot with ChatGPT can generate an account of the causes of the First World War or the principles of contract law. But knowing what to do about it, whether to trust it, how it connects to everything else, which bits matter and which are noise – these remain expensive, slow, human.

Judgement is not a skill you can look up. It’s a disposition you develop through practice – through getting things wrong and understanding why, through watching people who are better at it than you, through being held accountable by others who will tell you when you’re fooling yourself. AI can give us information. It can’t give us judgement.

Abundance makes it harder to know what we don’t know. When information was scarce, ignorance was obvious. Now, ignorance is invisible. We can generate confident-seeming text on any topic without understanding anything about it. The gap between performance and competence widens.

UCL’s Rose Luckin calls what’s needed “meta-intelligence” – not knowing things, but knowing how we know, knowing what we don’t know, and knowing how to find out. AI makes meta-intelligence more important, not less. If we can’t evaluate what the machine is giving us, we’re not using a tool. We’re being used by one.

That’s the equity issue that most AI boosterism ignores. If you went to a school that taught you to think, AI is a powerful amplifier. If you went to a school that taught you to comply, AI is a way of complying faster without ever developing the capacities that would let you do otherwise.

They didn’t quite succeed

Cultivating judgement means designing curricula around problems that don’t have predetermined answers – not case studies where students are expected to reach the “right” conclusion, but genuine dilemmas where reasonable people disagree. It means assessment that rewards the quality of reasoning, not just the correctness of conclusions – teachers who model uncertainty, who think out loud, who change their minds in public.

Creating communities of inquiry means spaces where people think together, are accountable to each other, and learn to be wrong in public. They can’t be scaled, and can’t be automated. They require presence, continuity, and trust built over time. AI can prepare us for these spaces. It can’t be one of them.

Last week I was playing with a custom GPT with a group of student reps. We’d loaded it with Codes of Practice and housing law guidance, and for the first time they understood their rights as tenants – not deeply, not expertly, but enough to know what questions to ask and where to push back. They’d never have encountered this stuff otherwise.

The custom GPT wasn’t the point – the curiosity it sparked was. They left wanting to know more, not less. That’s what democratised information synthesis can do when it’s not about producing outputs faster, but about opening doors others didn’t know existed.

Father, forgive me

There’s always been an irony in the complaint that graduates lack “soft skills”. For decades, employers demanded production – write the report, analyse the data, build the model. Universities obliged, orienting curricula around outputs and assessing students on their capacity to produce. Now that machines produce faster and cheaper, employers discover they wanted something else all along.

They call it “soft skills” or “emotional intelligence” or “communication”. What they mean is the capacity to be present with other humans. To listen, to learn, to adapt – to work with people who are different from you, and to contribute to collective endeavours rather than produce outputs in isolation.

It’s always irked me that they’re described as soft. They are the hardest skills to develop and the hardest to fake. They are also exactly what universities could have been cultivating all along – if anyone had been willing to name them and pay for them.

Universities that grasp this can offer students, employers and society something they genuinely need – people who can think, who can learn, who can work with others, who can handle complexity and uncertainty. Employers will need to train them in their specific context, but they’ll be worth training. That’s a different value proposition than “job-ready graduates” – and a more honest one.

I remember visiting the Saltire Centre at Glasgow Caledonian and being amazed that a university was brave enough to notice that students like studying together. Not just being taught together – studying together. The spaces that fill up fastest are the ones where people can work alongside others, help each other, and belong to something.

It’s not a distraction from learning. It is learning. The same is true of SUs, societies, volunteering, representation – the “extracurricular” activities that universities tolerate but rarely celebrate. These are where students practise collective action, navigate difference, take responsibility for something beyond themselves. Formation happens in community, not just in classrooms.

I tried not to do it

Being brave enough to confront all this will be hard. The funding model rewards efficiency, the regulatory model rewards measurability, and the labour market wants qualifications. The incentive is to produce – people who can perform, not people who have developed.

Students – many, not all – have internalised this logic. They want the degree, the credential, the signal. They are strategic, instrumental, and focused on outcomes. It’s not a character flaw – it’s a rational response to the system they’re in. If the degree is the point, then anything that gets you the degree efficiently is sensible. AI is just the latest efficiency tool.

But while shame is a powerful disincentive to the fess up, the thing about pretending is that it’s exhausting. And it’s lonely.

For years at Christmas, I pretended UWE was Bristol because I was ashamed – ashamed of wanting to study the media, ashamed of coming from a family where going to any university was exceptional, ashamed of the gap between where I was and where people felt I should be. The pretending was a way of managing the shame.

I suspect a lot of students feel something similar. The performance of knowledge, the strategic deployment of qualifications, the constant positioning and comparison – these are ways of managing the fear that you’re not good enough, that you’ll be found out, that the gap between who you are and who you’re supposed to be is too wide to bridge.

AI intensifies the fear for some – the terror that they’ll be caught, that the machine will be detected, that the pretending will be exposed. But it might offer a different possibility. If the pretending no longer works – if the performance can be automated and therefore has no value – then maybe the only thing left is to become someone who doesn’t need to pretend.

And I still don’t understand

That is the democratic promise of abundant information. Not that everyone will know everything – that’s neither possible nor desirable. But that knowledge can stop being a marker of status, a way of putting others down, or a resource to be hoarded. “I know something you don’t” can give way to “we can figure this out together.”

The shift from knowledge as possession to knowledge as practice is a shift from “I have information you lack” to “I can work with you on problems that matter.” From education as credentialing to education as formation. From “I’m better than you” to “I can contribute.” From pretending to becoming.

We’d need assessment that rewards contribution over reproduction. If the essay can be generated by AI, then the essay is testing the wrong thing. Assessment that requires students to think in real time, in dialogue, in response to genuine challenge – this is harder to automate and more valuable to develop. The individual student writing the individual essay marked by the individual academic is game over if AI can play both roles.

We’ll need pedagogy that prioritises encounter over transmission. Small group teaching. Sustained relationships between students and teachers. Curricula designed around problems rather than content coverage. Something between a module and a course, run by teams, with long-form purpose over a year rather than twelve-week fragments. Time and space for the slow work of formation.

We’ll need recognition that learning is social. Common spaces where students can study together. Student organisations supported rather than tolerated. Credit for service learning, for contribution to community, for the “extracurricular” activities where formation actually happens.

We’ll need slack in the system. The Swiss model, not the Czech one. Space to fall behind and catch up. Multiple attempts at assessment. Pass/fail options that encourage risk-taking. Time built in for things to go wrong, because things always go wrong. A system that absorbs delays rather than compounding them.

None of this will happen quickly. The funding model, the regulatory model, the labour market, the expectations students bring with them – they are not going to transform overnight. We’ll all have to play along for a while yet, doing the best we can within systems that reward the wrong things.

But playing along is not the same as believing. And knowing what we’re playing along with – knowing what we’re compromising and why – is the beginning of something different.

The end of pretending

The reason I came to work here at Wonkhe – and the whole point of my work with students’ unions over the years – has been about giving power away. Not hoarding insight, but spreading it. Not being the person who knows things – but helping other people act on what they now know.

The best email I got last week wasn’t someone telling me that I was impressive, or clever. I’ve learned how to get those emails. It was someone saying “really great notes and really great meeting – has got our brains whirring a lot.” Using what I offered to do something I couldn’t have done myself.

Maybe I’ve become one of those insufferable men who grab the mic to assert that what education is for is what it did for them. But the purpose of teaching is surely rousing curiosity and creating the conditions for people to become.

When I look back at the version of myself who told his family he was going to Bristol, I feel compassion more than embarrassment. He was doing the best he could in a system that made pretending rational.

Thirty years on, I’ve watched skills become obsolete, formats get put in the skip, pretences exposed. Each time we held on for a while. Each time we eventually let go. Each time something was revealed about what had actually mattered all along.

AI doesn’t end the system of pretending. But it does expose its contradictions in ways that might, eventually, make something better possible. If the performance of knowledge becomes worthless, then maybe actual formation – and the human encounters that produce it – can finally be valued.

The hopeful answer is that universities can be places where people become more fully human. Not because they acquire more information, or even because they become subject specialists – though many will – but because they develop the capacities for thought, action, connection and care that make a human life worth living.

They are capacities that can’t be downloaded, nor automated, nor faked. They can be developed only slowly, in relationship, through practice, with friction.

You came to university for skills and they turned out to be useless? That’s a trick. You came for skills and left ready to change the world? Now that’s magic.

Continue the conversation at The Secret Life of Students: Learning to be human in the age of AI – 17 March, London. Find out more and book.