Good Growth in the Age of AI: Institutions, constraints and the narrow path forward
Good growth is not simply about achieving a higher rate of GDP growth. It is growth built on sustained improvements in productivity that raise living standards, support social cohesion, and remain fiscally and environmentally sustainable. On that definition, the UK faces a profound challenge. Since the financial crisis of 2008, productivity growth has been weak by historical and international standards.
Despite repeated changes of prime minister and chancellor, little has altered this underlying picture. Public debate oscillates between resignation and exhortation. Some have quietly adjusted to slow, managed decline, while others continue to call for a decisive growth push without explaining how it might be achieved in a system whose core institutions appear overstretched, risk-averse, or captured.
In this feature article, I argued that Britain’s historical prosperity depended not only on trade, technology, or military power, but on an ability to build institutions that worked and that others chose to emulate. That institutional advantage has weakened. Today, many of our institutions struggle to deploy even existing technologies effectively, let alone absorb a new wave of powerful but imperfect tools such as generative AI. Without confronting this reality, there is a risk that AI becomes another layer of complexity added to fragile systems, rather than a source of sustained productivity growth.
Here I ask whether it is realistic to expect a new round of institution-led growth in the UK at all, and if so, what narrow path might still be open to a mid-power economy operating in a highly constrained and rapidly transforming political and geopolitical environment.
The binding constraint is institutional, not technological
On most measures, the UK sits downstream in the emerging AI stack. The core infrastructure and platforms – advanced semiconductors, hyperscale cloud, frontier foundation models – are concentrated in the United States and East Asia (particularly China). For a country in this position, competing primarily through tax concessions, planning relaxations, or subsidised energy risks entrenching a branch-plant model of the data economy, with high capital intensity, limited domestic capability, weak spillovers and little control.
At the same time, public service productivity, particularly in health care, has struggled to recover even to pre-pandemic levels. Growth remains concentrated in a small number of sectors and city centres, while many towns and regions depend on lower-productivity activity supported by underpowered local institutions. None of this reflects a shortage of algorithms or ideas. It reflects how work is organised, how services are designed, how risk is governed, and how decisions are made.
If institutions cannot use existing tools well, simply adding AI will not transform outcomes. Used poorly, AI risks reinforcing existing patterns and introducing new vulnerabilities, as systems become more complex, opaque and dependent on a narrow set of external providers. The constraint on good growth is institutional rather than technological. The more difficult question is whether that constraint can be eased without a major rupture.
Institutional change under constraint
History offers little comfort to those hoping for smooth, voluntary institutional renewal. Large-scale reform is rarely the product of tidy policy processes. It usually follows crisis – economic, political or geopolitical – and depends on unusually strong coalitions capable of overcoming entrenched interests.
China’s reform did not arise from a single moment, but unfolded through decades of gradual, uneven institutional change, much of it still unfinished. Even in a highly centralised system, reform has been slow, contested and partial.
The UK’s own major institutional reset in the late twentieth century similarly emerged from crisis. The inflation, industrial conflict and perceived national decline of the 1970s created a sense of urgency that made far-reaching change politically possible. That period unlocked a phase of growth, but it also entrenched new structural weaknesses that are now evident.
These examples point to an uncomfortable conclusion: without a significant shock, or a strongly constructed sense of crisis, the scope for deliberate, comprehensive institutional renewal is limited. Institutions tend to preserve the problems to which they are solutions. Reforming them from within is often more difficult than creating alternatives. Any discussion of good growth that ignores this reality risks drifting into wishful thinking. The challenge is to remain realistic about political and institutional limits while identifying what could plausibly move the needle.
What a good growth strategy can aim to do
A credible good growth strategy, under these conditions, should be understood as conditional and preparatory rather than programmatic. It cannot assume that government will suddenly acquire the appetite or capacity for long-term institutional investment. What it can do is more modest, but still demanding.
First, it can clarify the narrow path that might work for a mid-sized, open economy that does not control the frontier of key technologies.
Second, it can articulate institutional designs that could be adopted or scaled when political conditions allow, rather than improvising under pressure.
Third, it can guide incremental decisions taken now by regulators, public bodies, professional organisations and local actors, so that these moves at least point in a coherent direction rather than reinforcing drift.
Seen in this light, the role of the social sciences is not to offer yet another list of policy recommendations, but to help define what a viable institutional settlement for good growth would look like, and to keep that settlement intellectually alive.
Building alongside, not only reforming
One implication of this realism is that relying exclusively on reforming existing institutions is unlikely to be sufficient. Some are too overloaded, too constrained by accumulated compromises, or too exposed to short-term pressures to lead another wave of transformation. A more plausible approach is to build new institutional capacity alongside the old, rather than expecting wholesale replacement. This is not demolition for its own sake, but a form of institutional experimentation.
In health, for example, productivity has not returned to pre-pandemic levels despite rising demand and sustained funding. Alongside efforts to reform the NHS, it may be necessary to develop parallel models of care delivery built around new technologies, different funding mixes and clearer differentiation of services. If such models work, they attract patients and resources, forcing adaptation elsewhere. If they fail, the failure is contained.
The same logic applies to areas such as planning, infrastructure delivery and regulation. Creating parallel capacity allows learning and pressure without betting the entire system on unproven reforms.
Concentrating effort where leverage exists
Given limited institutional bandwidth, a good growth strategy must be selective. It should focus on domains where the UK still has dense ecosystems, reputational capital and plausible routes to scale.
One such domain is the tradable frontier: finance, insurance, professional and legal services, higher education, advanced manufacturing and parts of the creative economy. These sectors already operate globally, and some are adopting AI rapidly. The opportunity is not simply cost reduction, but the creation of new products, operating models and standards that can be sold internationally. That requires regulatory clarity, trusted governance and institutions capable of turning successful practices into durable norms at scale.
The other domain is the foundational economy: health and care, state education, transport, local government and basic utilities. These sectors place heavy demands on the public finances and shape everyday welfare. AI will only raise productivity here if it is accompanied by serious redesign of processes, roles and incentives. Even modest, compounding gains would ease fiscal pressure and support growth elsewhere.
In both cases, technology is important but secondary. What matters is whether institutions exist that can absorb, govern and scale change.
Finally, any serious good growth strategy must recognise the UK’s position in a fragmented, rapidly transforming global economy. With a few exceptions, the UK is unlikely to lead the technological frontier. Its real residual advantage lies in connectivity rather than scale.
The UK has deep ties to the United States, proximity to European regulatory regimes, longstanding professional and commercial links to India and the Commonwealth, and continued relevance to China through Hong Kong’s legal and financial infrastructure. Used well, this position allows the UK to act as a bridge between systems rather than as a proxy for any single bloc.
For AI and data-intensive services, this matters. Different technological and regulatory ecosystems are emerging. A mid-power that can offer interoperable rules, trusted governance and credible institutions has a chance to shape how these systems interact, even without controlling their core technologies.
A realistic good growth strategy
There is no easy path to good growth in the UK. Institutional renewal is hard, slow and often crisis-driven. Pretending otherwise does not help. Neither does resignation.
A realistic good growth strategy accepts political and institutional limits, concentrates effort where leverage exists, builds new capacity alongside the old, and uses the UK’s connective position in the global system to its advantage. It is not a promise of rapid transformation, but a disciplined attempt to keep a viable path open, so that when pressure for change intensifies, the choice is not between drift and improvisation. This is, ultimately, the most that a serious good growth agenda can offer.
About the author
Professor Feng Li is Associate Dean for Research & Innovation and Chair of Information Management at Bayes Business School. His research investigates how digital technologies can be used to facilitate strategic innovation and organisational transformation across different sectors and domains. He advises senior business leaders and policy makers on how to manage the transition to new technologies, new business models, and new organisational designs in the digital economy.
Photo credit: Mohamed Jamil Latrach on Unsplash