In recent years, artificial intelligence (AI) has leapt from research labs into mainstream discourse. Among the most visible developments is ChatGPT, a conversational AI that can draft text, answer questions, simulate dialogue, and even write code. For economists, this kind of AI is not just a novelty — it may be a harbinger of a structural shift in the drivers of economic growth.
In the UK, where productivity growth has long stagnated, the promise of AI as a new engine of expansion holds both hope and unease. Can ChatGPT and its successors reignite growth, foster innovation, and raise living standards? Or could they exacerbate inequality, erode jobs, and destabilise labour markets? In this commentary, I argue that the true impact of ChatGPT on economic growth will depend critically on public policy, institutional adaptation, and inclusive design.
This article is addressed to a UK audience: policymakers, business leaders, and the curious public. My goal is to explain the economics in accessible terms, highlight opportunities and risks, and suggest a roadmap for how Britain can harness AI for shared prosperity.
One of the enduring puzzles for the UK economy is weak productivity growth. Despite decades of investment in skills and infrastructure, output per worker has advanced sluggishly. Many economists lament a “productivity puzzle.” Could advanced AI like ChatGPT help break the logjam?
AI systems can automate or augment cognitive tasks — drafting reports, summarising research, answering customer queries, composing marketing copy, coding routines, and offering decision support. In principle, these systems can reduce “friction” in knowledge work, freeing up human labour to focus on innovation, strategy, and creative tasks. Over time, complementarities between AI and human labour might raise total factor productivity.
Some features of ChatGPT and similar large-language models are particularly suited to this transformation:
Scalability. Once trained, AI can serve many users at marginal cost close to zero. Its deployment can scale rapidly across sectors.
Adaptability. These models can learn domain-specific “fine-tuning,” making them useful across law, journalism, medicine, finance, and education.
Interactivity. Unlike batch automation, ChatGPT offers interactive dialogue, allowing users to refine and iterate tasks in real time.
Innovation feedback loop. Users may discover novel prompts or tasks for which AI becomes unexpectedly powerful, driving emergent use cases.
If these features can translate into measurable gains, then AI holds the promise of accelerating output growth beyond what conventional capital and labour accumulation could deliver.
To understand the mechanisms, consider the following channels through which ChatGPT-like AI could influence aggregate growth.
Labor Augmentation. AI may not replace all human work — instead, it can act as a powerful assistant, enhancing human productivity. A journalist might use it to draft an article, freeing time for interviewing or investigation. A lawyer might rely on it to summarise case law, leaving human oversight and argumentation to the expert.
Lower Entry Barriers. Startups and small firms often lack resources to hire large teams. Access to AI assistants can allow smaller players to compete more effectively, lowering fixed costs and promoting entrepreneurship.
Accelerated R&D & Innovation. AI can help researchers generate hypotheses, survey literature, propose experimental designs, and simulate outcomes faster. In effect, it may speed up the “innovation cycle” in science, technology, and product development.
Complementarity with Physical Capital. AI can improve the utilisation of machines, logistics, energy systems, and smart infrastructure by enabling more intelligent control, forecasting, and maintenance scheduling.
Spillovers & Diffusion. As AI tools become more widely adopted, knowledge spillovers may accelerate cross-industry productivity gains.
International Competitiveness. Countries that adopt advanced AI early may gain competitive advantages in high-value sectors, attracting investment, talent, and high-tech jobs.
In combination, these channels suggest that ChatGPT-style AI has the potential to shift the growth frontier outward, not just redistribute existing gains.
While the promise is alluring, significant risks loom. A sober economic assessment must reckon with labor dislocation, inequality, market concentration, misuse, and governance gaps.
One of the most immediate concerns is displacement. AI may automate routine cognitive tasks—such as drafting standard reports, repetitive content generation, legal or paralegal work, basic customer support. Workers in those domains might face job erosion or wage pressure.
The risk is not uniform: low-skill routine jobs are more vulnerable; high-skill, creative, or relational tasks are less so. This may deepen job polarisation: growth of high-skill, high-wage jobs at one end, and precarious low-wage work at the other, with hollowing out in the middle.
Further, labour markets may face adjustment frictions: retraining takes time, geographic mobility is costly, and social safety nets may lag. In the interim, many could face downward pressure on incomes or underemployment.
AI’s returns may accrue disproportionately to those who own the models, capital, and data — typically large tech firms and well-capitalised entities. This concentration could intensify income and wealth inequality.
Small firms or workers without digital access may be unable to benefit, exacerbating regional divergence (e.g., between more AI-enabled clusters and peripheral areas). The result may be winner-takes-most dynamics: the first movers dominate platforms, talent, and markets.
AI systems hinge on vast datasets. The quality, representativeness, and biases in these datasets matter immensely. Discrimination or exclusion in training data can embed prejudices — for example, in recruitment, credit scoring, or customer profiling.
Legal and ethical risks (privacy breaches, misuse of user data, disinformation, “hallucinated” factual errors) represent additional danger zones. A model may confidently assert false information, misleading users or causing reputational harm.
History warns against believing every technological revolution will deliver as promised. Prior waves (e.g. the dot-com bubble, biotech optimism) saw periods of hype that overshot fundamentals. The danger is capital may flow into “AI startups” with weak business models, leading to misallocation.
Moreover, if productivity gains fail to materialise fast enough, public disappointment may sour support, leading to regulatory backlash or diverging expectations.
AI development often outpaces regulation. Questions of liability (if AI makes harmful decisions), transparency (explainability of complex models), and accountability (who is responsible?) remain unsettled. Without robust governance, misuse, bias, or failures could undermine public trust.
International governance is also problematic: who sets standards for model safety, cross-border data flows, competitive spillovers, or “arms races” in AI development?
Training large models like GPT involves significant energy and carbon footprints. As models grow, so do their computational demands. If scaling AI is done without sustainable practices, environmental externalities could mount.
Because ChatGPT and its siblings are emergent technologies, empirical evidence is still nascent. But early case studies and experiments offer hints.
Business adoption experiments. Firms using generative AI for customer support, drafting, marketing, or internal knowledge systems report time savings and quality improvements.
Pilot studies in education. Some universities trial AI in writing feedback and tutoring; results are mixed — gains in efficiency, but concerns about overreliance and academic integrity.
Productivity metrics. Macro data have not yet shown a clear AI-driven inflection in growth across economies; any gains may be lagged or confined to narrow sectors.
Distributional patterns. Early adopters tend to be in tech, finance, media, or large organisations, rather than small firms or less digital sectors.
Model improvement curve. The performance of large language models has improved rapidly over successive versions, suggesting capacity for future leaps.
Together, these early signs hint at potential, but caution that diffusion, sectoral constraints, and institutional friction will matter a lot.
In the UK context, the impact of ChatGPT-style AI will depend on institutional conditions, sector structure, geographic disparities, and policy choices.
The UK has strengths in services, finance, media, professional services, and creative industries — domains where generative AI might be especially relevant. The challenge is to penetrate sectors that are more resistant (e.g. traditional manufacturing, construction, parts of healthcare) and to bridge the digital divide.
AI adoption may concentrate in London, Oxford-Cambridge corridor, Manchester, and other technology hubs. Regions with weaker infrastructure or human capital may lag farther behind, worsening the divergence between the South East and the “left-behind” areas. Addressing this requires policy coordination across regions.
The UK must invest heavily in digital literacy, lifelong learning, and reskilling. The ability of workers to collaborate with AI will matter — “prompt engineering,” critical thinking, oversight, and domain judgment will become more complemented by AI. Universities and vocational training systems need to adapt swiftly.
AI needs access to data, compute infrastructure, and robust connectivity. The UK must expand data infrastructure (e.g. cloud, edge computing, high-speed broadband) and policies on data sharing, while balancing privacy and security.
Britain has a comparative advantage in governance, ethics, and regulation: it could lead in setting responsible AI standards (akin to the UK’s leadership in fintech regulation). Clear rules on liability, explainability, fairness, transparency, and oversight will be essential.
Given the risk of unequal gains, the UK government must consider redistributive measures — progressive taxation, universal basic income experiments, wage insurance, or transition support — to ensure that AI-driven growth is shared rather than concentrated.
Public discourse and democratic oversight must be integral. The public should have a say in AI deployment in education, public services, media, surveillance, and social systems. Transparency, accountability, and institutional safeguards can prevent misuse or backlash.
To harness the promise of ChatGPT and mitigate the risks, here is a practical roadmap of policy initiatives the UK could adopt:
National AI Strategy with Growth Focus. Centre AI as a pillar of productivity policy, not merely a “tech agenda.” The strategy should coordinate research, industry support, infrastructure investments, and regulatory principles.
Public Sector “AI Labs.” Deploy ChatGPT-based tools in government, health, education, transport and justice systems to both improve public services and serve as testbeds for scaling and diffusion.
Data Trusts and Shared Infrastructure. Create frameworks for safe, privacy-aware data sharing between public and private actors, especially in sectors like healthcare, transport, environment, while giving citizens control over data use.
Regional Innovation Funds. Target funding and AI adoption programs in underserved areas, universities, and SMEs outside major tech hubs to close spatial gaps.
Lifelong Learning & Retraining Schemes. Expand adult learning grants, AI-augmented training programs, and modular credentials. Encourage “human + AI” curricula across universities, FE colleges, and bootcamps.
Responsible AI Regulation. Introduce sectoral regulation tailored to sensitive domains (e.g. finance, healthcare, justice), require transparency and auditability of AI systems, and define liability regimes for AI failures. Promote independent oversight bodies.
Tax & Redistribution Tools. Explore new taxation on AI rents, data dividends, or compute usage. Invest proceeds into social safety nets, worker adjustment, or public goods. Consider wage insurance or transition support for displaced workers.
Encourage Open & Cooperative AI. Promote open models, collaborative standards, interoperability, and public-sector provision of AI models to reduce lock-in and ensure competition. Encourage academia-industry public partnerships.
Ethics, Safety & Public Engagement. Establish citizen assemblies, ethics committees, and public consultation on the use of AI in public life. Fund research in AI safety, robustness, explainability, and bias mitigation.
International Cooperation. Collaborate with other governments on AI governance, cross-border standards, intellectual property, competition policy, and benchmarking. The UK should engage in EU, OECD, and global AI fora actively.
To make policy robust, it's useful to imagine adverse scenarios and stress-test assumptions.
Stagnant Productivity: AI fails to generate measured productivity gains beyond niches; businesses adopt cautiously, and expectations outpace reality.
Massive Job Displacement Without Safety Nets: Rapid automation displaces many middle-income jobs before reskilling can catch up, increasing unemployment and social strain.
Winner-Takes-All Tech Monopoly: A few firms consolidate AI infrastructure and application markets, thwarting competition and saddling the economy with dominance, stifled innovation, and rent extraction.
Regulatory Overreaction or Backlash: Public concern about AI misuse, false information, privacy, or job losses prompts overly restrictive regulation that stifles innovation.
Geopolitical Tech Rivalry: AI becomes a domain of strategic rivalry (e.g. between US, China, EU, UK), leading to fragmented standards, technology fencing, export controls, and reduced cooperation.
Ethical Failures or Bias Scandals: High-profile cases of AI bias, discrimination, misinformation, or hallucinations undermine trust, prompting public backlash and regulatory crackdowns.
Policymakers must build flexible, adaptive systems that can pivot if the trajectory diverges from optimistic forecasts.
ChatGPT and its successors represent a frontier in how economies can generate value from data, cognition, and interaction. For the UK, the stakes are high: successfully harnessed, generative AI could reignite productivity, foster innovation, and elevate Britain’s global competitiveness. Mismanaged, it could accentuate inequality, destabilise labour markets, and entrench dominance by a few large firms.
The key lesson is that AI is not destiny — its economic impact depends on choices: how we design institutions, regulate innovation, educate the workforce, and allocate gains. The UK, with strong traditions in governance, public institutions, and academic strengths, has a unique opportunity to lead — not just in building advanced AI, but in ensuring it yields broad-based prosperity.
I urge policymakers, business leaders, educators, and citizens to engage proactively. The technologies are evolving fast — but so too are the possibilities. Britain can steer AI toward growth that is inclusive, ethical, and sustainable. That must be our mission.
