91茄子

To survive the funding crisis, UK universities must embrace AI

The tasks that can be automated or augmented are some of those that university staff undertake day in and day out, says Alexander Iosad

June 20, 2025
A robot and office worker arm-in-arm, symbolising AI use
Source: PrettyVectors/iStock

The UK government’s recent White Paper on immigration, and its to student visas, have raised the spectre of a further squeeze on university finances. The proposed measures – such as reducing the duration of the graduate route visa from 24 to 18 months – are less radical than feared, prompting sighs of relief across the sector, but the warning lights are still flashing bright red.

University leaders will be acutely aware of the degree to which their budgets are propped up by income from overseas students, who constitute a quarter of the student population but pay 47 per cent of all tuition fees – and will be keeping an eye on the mooted levy on international fee income, among other things.

The government, on the other hand, faces a hard trade-off between two equally pressing policy objectives: reducing net migration and creating economic growth and opportunity through the UK’s world-leading universities.

New analysis by the Tony Blair Institute, , highlights that changes to student migration must come hand-in-hand with funding reform. Otherwise, we risk putting the sector, and the ?20 billion in annual exports it represents, at serious risk – particularly those parts of it that play the greatest role in social mobility.

91茄子

ADVERTISEMENT

But we are in an environment of great fiscal constraint: the state is spending more, taxing and borrowing more and yet failing to deliver better outcomes across the public sector. Some think this is a natural consequence of Baumol’s cost disease, whereby wage rises in more productive parts of the economy also buoy up salaries in low-productivity sectors – service-driven, labour-intensive industries such as higher education, in which meeting growing demand traditionally requires growing headcounts. Even standing still gets more expensive as the cost of inputs goes up. So the pressure builds, quality drops and sooner or later we reach the breaking point. We seem to be there now.

Of course, one way to address this would be to raise domestic fees, but this would do nothing to change the cost base or address the underlying factors that are pushing universities’ current operating model to the hard limit.

91茄子

ADVERTISEMENT

Yet there is now a new and plausible path forward – embracing the tools of artificial intelligence to reimagine how institutions work.

In truth, although AI has dominated public debate – or at the very least has taken over our LinkedIn feeds – over the past two years, universities have often seemed hesitant to engage with it. Such debate as there is has focused on the impact AI tools have in the classroom, often with a sharply negative inflection as lecturers struggle to adapt assessments to a world where telling apart work produced by students and chatbots is growing more and more difficult.

Yet across the public sector, AI offers the chance to break the hard link between quality, scale and headcount. Our analysis of tasks across hundreds of public sector bodies shows that, different as their remits are, AI has a common role to play in helping them engage with people who use their services, run their back-office operations and make better decisions.

We can reach out proactively to people based on what is known about their needs, circumstances or preferences. We can greatly streamline bureaucratic processes, automating compliance processes and triaging case work. Leaders can simulate the outcomes of multiple options in minutes based on real-time data and consult data banks of past failures and best practices.

91茄子

ADVERTISEMENT

Our analysis finds that existing AI tools alone could generate up to ?8 billion a year in productivity gains for local government and as much as ?40 billion across central government departments. The savings in higher education are likely to be significant too. After all, the tasks that AI can automate or augment are some of those that university staff undertake day in and day out: engaging with students, in or out of the classroom, current or prospective; dealing with admin and prioritising interventions to support staff or learners; making hard decisions to set and execute institutional strategies.

If anything, universities have even more reason than other service sectors to embrace AI given the impact these tools will have in accelerating research and personalising the learner experiences. The potential transformation here is immense. But it requires ambition, imagination and some degree of appetite for risk, including a willingness to invest now in future productivity gains: workflows will need to be rethought, job roles redefined, operations reimagined. This means hard choices, including potential job losses in some functions – but without this change, far more would be at risk.

No reform of universities’ funding settlement will change the structural deficit of their operating model – but embracing AI can deliver a significant reduction in the cost base and break the linear link between cost and quality. Few if any universities will thrive in the decade ahead unless they act now.

is director of government innovation policy at the Tony Blair Institute.

91茄子

ADVERTISEMENT

Register to continue

Why register?

  • Registration is free and only takes a moment
  • Once registered, you can read 3 articles a month
  • Sign up for our newsletter
Please
or
to read this article.

Related articles

The task of reading and rating the thousands of outputs submitted to the UK’s Research Excellence Framework is notoriously Herculean. Could AI ease the burden – or would its use undermine the whole point of having REF panels? As Jisc consults on that question, four writers offer their views

Reader's comments (9)

Ah but why "AI" rather than simply "IT". Universities were some of the early adopters of large IT systems for evrything from admissions to "virtual learning" to timetabling to management information. Much was spent on building, running rebuilding and scrapping these behemoths. As the bills piled up so did the understanding that many were simply getting in the way of staff and students doing their business. Yet like a conjuror promising to pull a hat out of the rabbit, here comes the author promising that this time it is really different, just pay up for my latest snake oil and youroperatonial problems and cost will be a thing of the past....
A major problem with all of this is that the time savings that are always promised with the adoption of new IT simply do not reliably materialise. They add labour at least as often as they save it. I am sceptical that AI will be any different because, frankly, current AI tools are not very good. They are so poor that everything they produce requires to be checked by a human.
I am not sure they are that poor and they are improving all the time (and this will be an exponential process), and anyway overall checking by "humans" is much less labour intensive than doing the work itself.
Well do you know, I am all for this AI malarkey! Bring it on!
" After all, the tasks that AI can automate or augment are some of those that university staff undertake day in and day out: engaging with students, in or out of the classroom, current or prospective; dealing with admin and prioritising interventions to support staff or learners; making hard decisions to set and execute institutional strategies". Rather vague on the whole. Chatbots? and "making hard decisions to set and execute institutional strategies"? Surely we don't want AI making any decisions hard or otherwise.
"workflows will need to be rethought, job roles redefined, operations reimagined. This means hard choices, including potential job losses in some functions – but without this change, far more would be at risk". Well I think we all know where Al Losad is going on this one. Or is Al actually AI?
This piece lacks concrete examples of how AI can be used. It's just handwaving.
Proprietary garbage in, garbage out.
new
Indeed, it would be great to see examples of such use. Sight of the protocol for the research that was undertaken to inform the piece would be of help as well. Any assumptions that were made for the analyses will help in making a judgment call and need to be communicated. Has there been any test cases for the claims made above?

Sponsored

Featured jobs

See all jobs
ADVERTISEMENT