Academia’s Legitimacy Crisis

Academia’s Legitimacy Crisis

Grade inflation, debt and generative Ai are devaluing the degree

SUMMARY
WHEN WRITING Goodbye Trust academia worried me the most. I realised the value of the degree was steadily eroding, but that’s happening faster than I had expected. Grade inflation weakens it as a marker of ability, effort and achievement. Accelerating levels of student debt means taking a degree is a significant financial risk: if the degree delivers less, does it justify the cost. Generative AI undermines confidence and reduces the value of the credential just as the graduate labour market decaying. Universities will continue to function, recruit students and award credentials, but this should not be mistaken for institutional stability. The danger is that the sector continues to survive while steadily haemorrhaging the legitimacy, authority and social value that once justified its privileged place in national life.

INTRODUCTION
The trust that academia delivers on its promises is eroding at a dangerous rate. Universities are not on the verge of collapse, although some may well need to close or merge. Most remain operationally stable although the accounts of their financial difficulties are becoming more frequent. In varying degrees they experience a cluster of pressures that weakens their legitimacy, authority and stability. A sector can continue to function despite the slow evaporation of what its credentials are supposed to signify, but only for so long.

The core diagnosis, set out in Goodbye Trust, was that academia was losing the trust of business and students while showing little appetite to confront the causes. Over the past year, that diagnosis has become more apparent. The pressures on higher education are becoming more tightly coupled. Generative AI weakens the link between visible academic output and underlying capability. Student finance is no longer a technical background issue, but an increasingly explicit argument about value, fairness, and repayment. At the same time, weaker labour-markets make it harder for newly minted graduates to find employment. My concerns about academia have heightened and this essay explains why.

THINGS ARE GETTING WORSE
The UCAS release on 28 January 2026 reported rising applicant numbers and a return to what it called ‘pre-pandemic’ patterns. The headline can be read as reassurance: 338,940 UK 18-year-olds had applied by the January deadline, up 4.8% year on year, with total applicants up 3.1%. That can look like proof that the degree bargain remains intact, that whatever people say about debt, grade inflation, or AI, the pipeline is still strong.

A closer read of the numbers suggest they show institutional momentum rather than restored confidence. The application rate among 18-year-olds remained broadly unchanged at 40.7% (in 2022 it 42.8%), applications from mature students declined, and international students continue to play an increasingly important financial role.

None of this disproves the argument of this essay. It shows, instead, how a system can remain busy even as the reasons for trusting it decline. Demand can be sustained by demographics, habit, and cross-subsidy, from high fees paid by overseas students, long after confidence in the underlying promise has begun to wobble.

When writing Goodbye Trust, academia was the institution that worried me most, not simply because its obvious failing but because it possessed a dangerous kind of resilience. Unlike business, it does not face direct market discipline. Unlike politics, it can absorb public dissatisfaction for long periods without immediate correction. Standards can weaken, diversity of thought can narrow, credential credibility can fall yet the institution’s response is to claim its insulation from external pressures as being necessary for retaining its independence.

cess students and produce credentials while slowly debasing their value. A degree is not simply a learning experience. It is a social promise, a claim about effort, competence, and readiness that employers, students, and governments are expected to trust. If that claim weakens, the system need not collapse to fail. It can continue operating while strain becomes more evident in course closures, hiring freezes, mergers, and increasing reliance on Chinese students.

A year ago, it still seemed that the drift would be slow enough for institutions to adjust incentives, restore clarity in assessment, and re-earn confidence. That now looks optimistic. The pressures are moving faster than the machinery of universities is designed to handle, and they are increasingly reinforcing one another. The argument that follows is therefore not a fresh verdict on universities so much as an attempt to show why their time margin for adjustment is fast disappearing.

LEGITIMACY DECLINES
Before turning to the newer shocks, it is worth establishing the baseline. Many of the concerns now attached to universities did not begin with AI, the pandemic, or the latest controversy over student finance. They were visible earlier, developing slowly and unevenly, and were often dismissed as noise rather than signal.

One of the clearest indicators is the changing distribution of degree classifications. Over recent decades, the proportion of students awarded first-class degrees and upper seconds has risen sharply, to the point where what was once exceptional has become common. The share of firsts rose from single digits in the mid-1990s to roughly a third by the early 2020s, and the combined share of firsts and 2:1s now sits above three-quarters.

What makes that pattern awkward is that it coincides with the mass expansion of higher education. Participation has widened, entry requirements have loosened and yet top classifications have continued to rise. Reporting has repeatedly noted that especially large jumps in firsts have often been concentrated in lower-ranking institutions. This does not prove that all those results are undeserved. It does show why the system-level claim that this is simply better teaching everywhere is patently wrong.

The Office for Students has acknowledged this problem directly. Its modelling suggests that a substantial share of the growth in top classifications cannot be explained by changes in student intake, subject mix, or demographics. Regulators are careful not to turn that finding into a blanket claim that standards have fallen, and rightly so. But from the perspective of signalling, the distinction is less consoling than it first appears. When many more students cluster at the top of the scale, the value of classifications weakens.

The mechanics of classification create their own upward pressure. Discounting low marks, weighting later years more heavily, expanding resits and resubmissions, replacing exams with coursework, and using ‘best-of’ algorithms may all be defensible in isolation. Taken together, they tend to push in one direction. The question is no longer whether any single rule is illegitimate, but whether the cumulative effect is a grading system that is not fit for purpose.

Why this matters is that degree classifications are not merely records of achievement; they are social signals. Employers use them as screening devices. Professional bodies use them to regulate entry. Students use them to justify years of study and substantial debt. When the signal becomes noisier, alternative proof of performance becomes more important. The value of the university brand, work experience, internships, networks, psychometrics, and student portfolios. Each additional filter reduces the degree’s standalone value.

A parallel strain sits beneath this. The financing model is no longer just a technical design problem; it has become part of the legitimacy story. Britain has an enormous student-loan book, over a quarter of trillion pounds. The terms of its repayment are increasingly experienced not as a stable contract but as a politically adjustable levy. Student finance is moving from a specialist argument about mechanics to a broader dispute about fairness, value for money, and who and when the losses are absorbed.

The point is not that merit has vanished. It is that the system enters the current moment with thinner margins of credibility than it once had. Grade inflation and a politicised loan model are not, by themselves, terminal. They make new shocks harder to absorb.

THE DEGREE CONTRACT WEAKENS
For years, the real asset behind mass higher education was not primarily financial. It was the degree’s social authority. For much of the middle class, university acquired the quality of inevitability, a career route one had to actively refuse. Parents expected it, schools channelled it, friends disappeared in September, and staying behind could feel like opting out of the future. To have a degree was not simply to be educated, but to be legible as someone still ‘on track’.

That sense of inevitability insulated the system from value-for-money questions. So long as ‘everyone goes’ remained the cultural rule, universities did not have to persuade; they had to look appealing and process the applications. But inevitability is not a law of nature. Mass participation is historically recent and politically contingent. A system that expanded at speed can contract at speed once the default story shifts from ‘this is what people like us do’ to ‘why are we doing this at all?’

The shift matters because belief changes before aggregate demand does. Within families, younger siblings do not need a policy paper to reassess the bargain. They watch older brothers and sisters struggle to translate degrees into secure work while carrying enormous debt. If the first graduate in a family ends up underemployed and resentful about loans, that experience becomes a more powerful argument than any official reassurance. This is how a legitimacy problem turns into a demand problem: belief updates at home before it appears in the statistics. However, never underestimate the power of post‑purchase rationalisation. It’s hard for young people to admit to making a mistake – ‘OK I haven’t got a job but it was the best time of my life – made lots of friends – got to travel the world etc etc.’

At the same time, the degree’s signalling power is weakening from several directions at once. Mass participation dilutes exclusivity. Grade inflation corrodes trust in classification. Generative AI adds a new form of suspicion, not that every student cheats, but that the outsourcing of academic performance devalues the credential. A qualification that cannot reliably signal effort and competence ceases to function as a trusted filter.

That lands directly on the individual calculation. Large debts feel less tolerable when the earnings uplift is uncertain, when graduate jobs are harder to secure, and when the earlycareer rungs are highly competitive. Meanwhile, the lived experience of university has, for many students, been quietly downgraded by housing costs, remote learning, and the growth of commuting from home. The premium product increasingly looks, for a sizeable share of the market, like paying more for a diminished experience and a weaker signal. The evidence does not show that higher education has lost all economic value, nor that graduates as a class are simply worse off than non-graduates. Returns still vary sharply by subject, institution, and subject. What has changed is the dispersion of benefits with more students occupying the lower-return tail of the distribution, which weakens the reliability of the promise at the point of enrolment. The likely result is not a clean collapse but a re-sorting: a more class-coded system in which the affluent retain the fuller residential experience and its social capital, while the mass market receives a diminished product with a weaker signal of value.

FRAGILITY OF STUDENT LOANS
I will continue to use the term ‘student loan’ despite evidence indicating it is as much a grant as a loan. Only 56% of full-time undergraduate students, starting in 2024/25, are forecast to repay their loans in full. This is an increase over the 2022/23 cohort, where just 32% were expected to complete repayments.

Higher-education finance determines who bears the risk when expectations are not met, how losses are absorbed, and how long dissatisfaction can be deferred before it turns into pressure for change. In England, that deferral has reached a scale that matters not only politically but fiscally.

The current model rests on two linked mechanisms: a tuition fee capped in cash terms for much of the past decade, and an income-contingent loan system that substitutes graduate repayment for direct public funding over long periods. Together, these arrangements allowed for expansion while keeping headline public spending lower than it would otherwise have been. Costs have been pushed into the future turning higher-education finance into a longdated fiscal exposure rather than a conventional current-budget item.

That exposure is now explicit. Outstanding student-loan balances in England were approximately £267 billion in 2024–25, that’s four time the annual spending on defence. Official projections suggest the loan book could approach £500 billion in real terms by the late 2040s, before write-offs peak.

For most borrowers, compliance with the rules means watching their nominal loan balance grow rather than decrease.

Income-contingent loans were meant to protect lower earners and smooth repayment across the life course. In that respect, they work as designed. But the same design ensures that cost is deferred rather than removed. For many graduates, additional interest does not increase what they will actually repay; it increases what will eventually be written off. From the state’s perspective, the asset’s economic value is materially below its face value.

This issue has now become more politically visible. On 7 April 2026, the government announced a one-year 6% cap on the maximum student-loan interest rate from September 2026, affecting some higher-earning Plan 2 borrowers and those with postgraduate loans. The move matters less for what it saves than for what it reveals. It does not reduce monthly repayments, which remain tied to earnings rather than to the outstanding balance. For many borrowers, especially those unlikely to repay in full, the practical effect is mainly to reduce the amount eventually written off rather than the amount they themselves will pay. In other words, the policy acknowledges pressure around the optics and politics of loan balances without resolving the deeper problem of legitimacy. If you were looking for an instance of ‘kicking the can down the road’, student loans are the perfect example.

That is where the issue moves beyond graduate grievance and into fiscal constraint. Student loans affect the presentation of the public finances, and the timing of debt recognition matters politically. For years, that complexity helped keep the issue at arm’s length from mainstream debate. That insulation is weakening. Once repayment is experienced as a longduration levy attached to a credential whose value is being openly contested, it will continue to become a news item. Universities depend not only on cash flow but on stability of expectations, and politicisation destabilises both.

Despite the enormous amount of student debt the finances of universities are fragile. Repeated fee freezes during periods of rising costs have reduced real funding per student over the past decade. Universities have responded through efficiency drives, cross-subsidy, and non-core income, but these responses have reduced margins and the system’s resilience. Public argument, therefore, tends to focus on the wrong level. Debates about whether graduates were mis-sold a product or charged too much interest are not trivial, but they miss the larger question of what happens when the state is forced to acknowledge that a substantial share of a very large asset will never be repaid and that the survival of many universities will require state funding, at a time when the state’s finances are in a dire position.

Seen in that light, more radical proposals become easier to understand. When the Greens argue for abolishing tuition fees, restoring maintenance grants, and cancelling existing student debt, they are not only making a fairness claim. They are, implicitly, proposing earlier fiscal recognition: stop pretending the loan book is a recoverable asset and fund higher education transparently through taxation instead. Labour’s more managerial approach, adjusting thresholds or terms at the margin, may relieve pressure without altering the underlying exposure.

The longer that recognition is deferred, the more likely policy becomes a sequence of tactical adjustments, each intended to ease pressure without confronting the larger question of what higher education should cost and who should bear that cost.

A DANGEROUS FEEDBACK LOOP
Any one of these pressures might be manageable in isolation. What makes the present moment different is the way they now interact. Weaker labour-market demand undermines the perceived value of the degree just as loan burdens become more salient. Grade inflation weakens signalling, which generative AI weakens further by making academic performance easier to simulate. Employers respond by looking for other signals than the degree creden tial. Early-career disappointment then feeds back into repayment capacity and political tolerance for deferred costs. The loops are becoming tighter and more direct.

Universities have, in effect, already conceded the central point about assessment. The question is no longer whether students use AI, nor whether software can reliably detect that use after the event. Across the sector, the more serious recognition is that traditional coursework no longer provides secure evidence of independent thought. Once that is admitted, the problem changes. It is no longer about policing misconduct; it is about reconstructing credibility.

So far the responses are tactical more than strategic: more in-person exams, more oral assessment, more draft submission, more reflective commentary, more version tracking, and formally declared AI use within specified limits. Their shared logic is plain enough. If final output can no longer be trusted on its own, institutions have to seek signs of process, provenance, or live performance that are harder to outsource or simulate. But adaptation should not be mistaken for resolution. Oral assessment is expensive, process tracking can itself be gamed, and a return to tightly controlled examination looks less like institutional confidence than defensive retrenchment.

Financial pressure makes that redesign harder. Universities with squeezed margins have limited capacity to rebuild assessment at scale, invest in credible verification, or absorb the reputational cost of admitting how far older methods have weakened. Stability therefore becomes a strategy, but a brittle one: preserve the surface for as long as possible even as the substance becomes less secure. A further complication is access. Any broad move back toward tightly controlled assessment has to contend with a large, partly unseen population of students with disabilities or other needs for accommodation. The design problem is not merely how to restore credibility, but whether credibility and accessibility can still be designed together at scale. When about 20% of students have some sort of disability (mainly mental), then the scope for increasing pressure on them is limited. The same argument applies to students whose first language is not English.

Alongside these pressures sits another source of institutional brittleness. Over the past decade, many universities have embedded DEI frameworks across governance, hiring, promotion, assessment, and student management. The relevant point here is not whether every such initiative is misguided. It is that frameworks tied to compliance, reporting, and reputational risk can narrow the space for internal challenge. When staff believe disagreement carries professional risk, self-censorship becomes rational. When hiring or promotions are seen to reward alignment alongside merit, confidence in standards weakens further.

In a system whose legitimacy is already threatened, any framework that makes internal criticism harder or experimentation riskier slows learning precisely when institutional agility is needed. Combined with financial stress, labour-market pressure, and assessment uncertainty, this produces not a dramatic collapse but institutional brittleness. All in all, this paints a bleak picture of academia.

HOW UNIVERSITIES FAIL
Institutional failure is depicted as a total collapse: empty buildings, mass redundancies, and the sudden withdrawal of support. That is unlikely to be the default outcome for higher education while the current settlement broadly holds. Universities are too embedded, too regulated, and too socially entangled for traumatic failure. Degrees continue to be awarded, research continues to be produced, and campuses remain busy, but the institutional foundations destabilise.

This pattern persists because universities are practised at muddling. Despite the domestic demand remaining resilient the political atmosphere is shifting. What once sat in the background as a deferral mechanism is becoming a visible legitimacy issue, framed as mis-selling, or intergenerational unfairness. So long as the settlement survives, degradation can continue. A decisive rupture, however, would change the failure mode. Sharp fiscal tightening, a major reduction in overseas student income, or political intervention that materially changes loan finance, fees, or student numbers could turn slow erosion into rapid contraction.

That distinction helps explain current institutional behaviour. Universities default to caution not because they cannot see the problems, but because continuity still rewards it. Yet the margin for gradual correction is narrowing. A weak economy can compress graduate returns even if universities behave well, and historical statistics on the graduate premium describe an older world in which universities could still claim a stronger monopoly over certified cognitive performance. Generative AI weakens that monopoly. If employers increasingly price skills and AI-augmented performance more directly, the degree becomes a slower and less reliable proxy, and the behavioural shift will arrive before the official data fully records it.

There is also a harder possibility to confront. Universities do not merely respond to the economy; over time they help shape the cohort that enters it, the mix of skills, incentives, and institutional habits that influences productivity and risk-taking. That makes the economy and higher education less separable than they often appear in public argument. The danger, then, is not simply that universities suddenly stop functioning. It is that they continue functioning under conditions that progressively weaken trust until correction is imposed from outside, on harsher terms and on a tighter timetable than the sector would ever have chosen for itself.