Exploring speculative scenarios – however unlikely they may seem – allows us to envision the future we want (or perhaps want to avoid). In the process, we can examine the role of our institutions – and the people who work in them – in planning for the future. In this piece, we visit a higher education campus in the UK, five years from now. This article first appeared on the UNESCO website.
Scenario location: A university town in England. Year: 2030.
Professor Eleanor Wright is walking through the sunny campus. It’s Friday – her teaching day. The other four days of her working week are spent at the offices of the AI lab in the big city, an hour away. Her university contract requires six hours of physical presence per week, primarily for teaching undergraduates and delivering a prestigious lecture series.
As she passes ancient buildings, she reflects on how her career has transformed over the past few years. As a tech-savvy researcher with experience of managing big projects, a strong publication record and a string of national media appearances, her colleagues called her a ‘superstar academic’. Most of these colleagues are now gone. As AI models continued to improve and improve – nobody quite knows when the breakthrough was, if there even was one – the power of the big AI labs grew too. Their hunger for new knowledge and their immense computing power muscled universities out of the way. Cash strapped after years of funding crises, only the superstars managed to hold on to their jobs.
The title ‘professor’ remains coveted, but its meaning has fundamentally changed – her primary value to the university now lies not in the research she performs within its walls (which is negligible), but in the funding her secondment brings. Her reputation, as the face of the human-machine research team at the AI lab, also helps attract undergraduates. The company that eviscerated her university has made her more famous than her appearance as a BBC panelist, she reflects wryly.
The great migration
Back in 2025 there was talk of an ‘intelligence explosion’ that would fundamentally reshape knowledge creation. This turned out to be true, sort of. For the average bystander, the explosion was gradual and quiet. After 2025, the migration of elite researchers from universities to industry-funded AI laboratories accelerated, creating what some have termed the “great academic migration”. Many (non-superstar) academics protested, but university leadership teams played along. Whether they were pragmatic visionaries or worn down by decades of under-investment depends on who you ask. But framing this migration as a series of secondments offered leaders a fig leaf of self-respect, and gave the labs what they needed most: the validation and legitimacy of the storied institutions they were overtaking; a badge to attach to their superior computational resources.
This migration played out unevenly across the globe. In high-income countries, the pattern followed the UK’s example, with elite research universities transformed into teaching-focused institutions with a handful of superstar researchers. In emerging economies, particularly in parts of Africa and Asia, national governments intervened more forcefully, creating public-private partnerships that preserved more university autonomy while still accessing AI lab resources. The resulting global knowledge landscape became more complex, with research hubs distributed according to the decisions of AI labs rather than university prestige.
The evolution of research
The prevailing narrative that this represents a catastrophic loss for universities requires nuance. Most institutions survived this transition, but are transformed. What we are witnessing is not the end of the university but rather a reformulation of its purpose. Universities are centres of teaching excellence, of critical enquiry, of engagement with small businesses. Cutting-edge research has mostly moved to the AI labs – shiny university research parks are now student accommodation blocks and seminar rooms.
The nature of research itself has also fundamentally changed. ‘AI scientists’ – autonomous systems capable of formulating hypotheses, designing and conducting experiments, and publishing findings – evolved from prototypes to mainstream research tools. These systems were first developed in the mid-2020s as open-source projects, and could produce publication-quality papers for the cost of a pizza and share findings amongst themselves. They were later refined by every major AI laboratory, accelerating research productivity in fields amenable to computational methods. The explosion of research papers predicted by early commentators did indeed materialise, though the feared quality crisis was largely averted through sophisticated review systems.
The relationship between Professor Wright and her AI hosts illustrates the new academic compact. She provides contextual expertise, ethical oversight, and academic credibility; the company provides computational resources that would bankrupt a university department. The research outputs carry both her name (and university affiliation) and those of the lab technicians, and are published in open-access journals. The university receives a generous annual payment (balancing its books); the company retains the IP and commercialisation rights (boosting its share price). This arrangement is the standard model for frontier research across disciplines from theoretical physics to computational sociology.
Resilience and resistance
We should be careful, however, not to overstate the completeness of this transition. Some research domains remain stubbornly resistant to the magic of the AI labs, particularly those requiring extensive human involvement or tacit knowledge. The arts and humanities have proven especially resilient and are seeing something of a resurgence amongst undergraduates. What we have seen is not a wholesale replacement of human researchers, but rather a redistribution of cognitive labour, with AI systems taking on increasingly sophisticated components of the research process. In parallel, human ‘taste’ is highly prized – the market value of human artwork continues to shatter records.
The implications for universities are far-reaching, and many are still catching up. The teaching-research nexus – the idea that university education should be enriched by active research – has been reconceptualised. Rather than teaching their own research, academic staff now teach the process of engaging critically with a rapidly expanding knowledge frontier, mostly produced outside university campuses. Students learn not just established knowledge but how to navigate, evaluate, and contribute to an information system dominated by AI-human research teams. This required new pedagogical approaches focused on developing what might be termed “second-order knowledge” – not merely understanding content but understanding the systems through which knowledge is produced, validated, and disseminated.
The future of universities
This transformation continues to raise urgent questions about the future of universities. If their research function continues to diminish, what justifies their privileged position in society? The answer lies partly in their role as custodians of methodological rigour and critical thinking in an information environment increasingly dominated by commercial interests. Universities have become essential counterweights to corporate knowledge production, training students to interrogate the biases, limitations, and ethical implications of AI-generated research – but there’s a sense this counterweight isn’t quite heavy enough.
When asked to comment on the string of breakthroughs now taking place in AI labs, their CEOs are quick to praise universities. Professor Wright’s weekly journey between university and laboratory represents not the decline of academic tradition but its evolution, they say – a recognition that knowledge creation has always adapted to the technological and institutional constraints of its time. Professor Wright is less certain. As she enters her office, she sees only the ghosts of her former colleagues and coworkers.
(Cover image: Datafication by Kathryn Conrad, via Better Images of AI)
Leave a comment