The Cognition Economy

There haven't been many quarters in the last three years that have ended without the release of a new AI model. The degree to which each iteration has genuinely advanced us is debatable, but one thing is not: at some point in the not-too-distant future, these models will equip organisations with the ability to produce analyst-level solutions across almost every discipline.

The most eager businesses, recognising what is to come, have already begun acting. Some have frozen graduate hiring, while others have gone as far as disposing of tenured employees after automating the more technical aspects of their roles.

"Improving efficiency", "building leaner organisations", creating more agile teams and freeing up capital to invest in cutting-edge innovation have all been cited as reasons for these decisions. While these explanations give the impression that businesses are positioning themselves to thrive in the age of AI, they also reveal a failure among decision-makers to view technology as just one (albeit an incredibly important) component of a broader strategy for sustaining competitive advantage.

When firms prioritise AI adoption without considering how it reshapes the way they differentiate themselves, they risk entering a race towards homogeneity, where every product, service and strategic recommendation increasingly resembles the next. Economic theory tells us that homogeneity rarely commands a premium. Consumers and enterprises alike are willing to pay more for products distinguished by quality, originality or scarcity - not those that simply match what everyone else can produce.

The deeper shift, however, may not concern products at all. It concerns how organisations have historically produced thinking itself.

For decades, graduate schemes and structured early-career pathways have functioned as more than training programmes. They have served as mechanisms through which organisations cultivate cognitive cohesion. Beyond teaching employees what to do, they shape how problems are framed, establish acceptable ways of reducing ambiguity and reinforce particular approaches to decision-making. In doing so, these programmes have reliably produced employees who progress into middle management because they are dependable, legible and aligned with the organisation's prevailing way of thinking.

This process of organisational socialisation has long been recognised within management literature. Chang, Wannamakok and Hsi (2023) argue that organisations which train, reward and promote employees primarily for their ability to replicate established systems often do so at the expense of long-term innovative thinking. Likewise, Van Maanen and Schein suggest that organisations reproduce themselves by encouraging conformity to institutional norms and behaviours. While this creates consistency and operational efficiency, it can also diminish the forms of cognitive divergence that enable organisations to differentiate themselves over time.

Ironically, organisations are now replacing one system designed to produce cognitively homogeneous analysis with another.

Human graduate analyst out. Digital graduate analyst in.

Despite being trained through fundamentally different mechanisms, frontier AI models are increasingly optimised for remarkably similar outcomes. They are rewarded for generating responses that are coherent, plausible and aligned with the knowledge on which they were trained. Like the graduate analyst, they are incentivised to produce the most defensible answer rather than challenge the assumptions embedded within the question itself. Different systems, remarkably similar optimisation functions.

This is not a criticism of either approach. There is considerable value in consistency, and much of AI's commercial appeal lies precisely in its ability to produce reliable, high-quality analysis at unprecedented speed and scale. The problem emerges when analytical capability itself becomes abundant. If every organisation has access to similarly capable models, and those models increasingly replace cohorts of employees who were themselves trained to think in similar ways, organisations risk converging on increasingly similar conclusions.

The consequences of this shift are already beginning to emerge in the labour market. Between February and April 2026, youth unemployment in the United Kingdom rose to 16.2%, with 735,000 people aged 16 to 24 unemployed. It would be simplistic to attribute this entirely to artificial intelligence. Economic conditions, hiring cycles and broader market uncertainty all play an important role. Yet as organisations automate more entry-level analytical work, a more fundamental question begins to emerge.

What happens when businesses stop manufacturing judgement?

Graduate programmes have historically served not only as recruitment pipelines but also as apprenticeships in organisational reasoning. They were environments in which employees accumulated tacit knowledge, developed commercial intuition and learned to exercise judgement through repeated exposure to increasingly complex decisions. If those pathways continue to contract, organisations lose more than junior analysts. They lose one of the primary mechanisms through which managerial judgement has traditionally been cultivated.

Paradoxically, the removal of these systems may also create the conditions for a different kind of competitive advantage.

Rather than producing successive generations of cognitively similar managers, organisations may increasingly depend upon individuals whose value lies precisely in thinking differently. Individuals who move between industries, disciplines and organisations accumulate perspectives that cannot easily be reproduced through either institutional socialisation or probabilistic models. Their advantage lies not simply in possessing knowledge, but in connecting ideas that rarely meet.

The signals are already visible at the individual level. In conversations among recent graduates, a pattern is emerging that would have seemed unusual a decade ago: young people actively seeking exposure across product, strategy, marketing and operations within the first years of their careers - not because they cannot commit to a specialism, but because they have correctly identified that breadth is itself becoming a form of capital. The portfolio career, once associated with freelancers and those between roles, is being reframed as a deliberate strategy. Project-based engagements across different industries, each one adding a different problem-solving vocabulary, are beginning to look less like instability and more like the most rational response to an economy that will reward the ability to connect ideas that institutional training tends to keep separate.

This reframing matters because it changes what a career is for. The linear progression - analyst to associate to manager to director - was always partly a developmental structure, a way of accumulating judgement through increasing exposure to consequence. If that structure is contracting, individuals who understand what it was actually producing will find other ways to produce it. Those who do not may find themselves five years in with strong technical credentials and a narrower base of experience than the economy increasingly demands.

The harder question falls on organisations. If firms stop hiring at entry level and begin recruiting at the five-to-seven year mark - bringing in people who are already formed - they are making an implicit assumption: that the developmental infrastructure they have dismantled still exists somewhere else. That assumption may not hold for long. As graduate programmes contract across industries simultaneously, the pipeline of experienced, judgement-capable candidates does not simply persist on its own. It requires investment somewhere. The organisations that stop making that investment are, in effect, free-riding on a system they are collectively eroding.

What fills the gap is not yet clear, but the shape of it is beginning to emerge. Some organisations may reconstruct something resembling what they retired - short-term project deployments, structured rotations, cohort-based engagements that serve a developmental function even when they do not carry the name or the cost of a traditional graduate scheme. Others may outsource that function entirely, relying on a new category of firm: consulting teams built not around sector expertise but around the deliberate development and deployment of judgement at scale. The deployment model, in this reading, is not simply a flexible staffing arrangement. It is an institutional response to a capability gap that organisations created but have not yet acknowledged.

In either case, the underlying dynamic is the same. Judgement — the ability to synthesise across disciplines, to identify what matters in conditions of ambiguity, to make calls that models cannot — is becoming both more valuable and more difficult to manufacture. The organisations that recognised this early and continued to invest in the conditions that produce it will be better placed than those that treated the question as someone else's problem.

The cognition economy is not defined by intelligence becoming artificial. It is defined by judgement becoming scarce.